# Falcon docs, every page

This file joins every page of https://falcon.com/docs. The index is https://falcon.com/docs/llms.txt.

---

Source: https://falcon.com/docs

# Falcon docs

In Falcon, everyone who knows a business shapes the same AI agents and workflows, and the agents pass work to each other.

You shape each agent in plain words. Several people can work on the same agent, and nothing reaches customers until someone makes a version Live. Agents run on Amazon Bedrock AgentCore, in Falcon’s AWS account or your own. Everything you can do in Falcon is also available to AI assistants over MCP.

Falcon is in a private pilot. Some features are live, some are being tried with pilot companies, and some are still to come. Each feature says which.

## Getting started

1. [Sign in and set up your workspace](https://falcon.com/docs/account): Sign in with a code sent to your email. The first administrator claims the workspace and chooses how people sign in.
2. [Invite your team](https://falcon.com/docs/company-settings#people): Invite the people who will build and look after agents, and give each a role.
3. [Build your first agent](https://falcon.com/docs/agents): Start from a starter or a description. Write its instructions, then add knowledge and tools.
4. [Add the people and places it speaks as](https://falcon.com/docs/identities): Add an identity for each person or place the agent acts for, with their details and hours.
5. [Connect your apps and channels](https://falcon.com/docs/connections): Connect the apps, mailboxes and numbers the agent needs, and choose where customers can reach it.
6. [Test it, make it Live, and add a workflow](https://falcon.com/docs/workflows): Talk to the draft, make a version Live, then start it from the events that matter to you.

## Guides

- [Sign-in and your account](https://falcon.com/docs/account.md): How people join their company in Falcon, sign in without a password and look after their own settings.
- [Agents](https://falcon.com/docs/agents.md): Build an agent in plain words, give it skills, knowledge and tools, test it, then make a version Live.
- [Identities](https://falcon.com/docs/identities.md): The people and places your agents speak as, each with its own details, hours and ways to be reached.
- [Workflows](https://falcon.com/docs/workflows.md): Chain agents and steps together, start them from events, and decide what happens when someone replies.
- [Connections](https://falcon.com/docs/connections.md): Connect the apps, mailboxes, numbers and storage your agents and workflows use.
- [Company settings](https://falcon.com/docs/company-settings.md): People, roles, security, models and spend limits, billing, the audit log and your data.
- [Reports](https://falcon.com/docs/reports.md): Counts and costs for agents and workflows, never what people said.
- [Ask Falcon](https://falcon.com/docs/ask-studio.md): An assistant on every page that explains what you are looking at and can make changes for you, with your say-so.
- [MCP and API](https://falcon.com/docs/mcp-and-api.md): Everything you can do in Falcon, an AI assistant or your own code can do too, within your access.

## Reference

- [Every feature](https://falcon.com/docs/features.md): 107 features: 0 live, 90 in pilot and 17 coming next.
- [API and MCP reference](https://falcon.com/docs/api.md): every operation an assistant or your own code can call, grouped by area.
- [Glossary](https://falcon.com/docs/glossary.md): The words Falcon uses, in plain language, in alphabetical order.
- [Choose how to run Falcon](https://falcon.com/docs/choose-how-to-run.md): Run your agents with Hosted by Falcon, or in your company's own AWS account. You build them the same way in both.
- [Connect your assistant](https://falcon.com/docs/connect-your-assistant.md): Use Falcon from Claude, ChatGPT, Cursor, VS Code or another AI assistant over MCP. It signs in as you and can do only what you can do.
- [Worked examples](https://falcon.com/docs/examples.md): Real requests and payloads for webhooks out, request maps and the failure handler, ready to copy.
- [Troubleshooting](https://falcon.com/docs/troubleshooting.md): What each problem code means and what to do about it, with the exact title the error shows.

---

Source: https://falcon.com/docs/account

# Sign-in and your account

How people join their company in Falcon, sign in without a password and look after their own settings.

Falcon has no passwords. You sign in with a code sent to your email, and you can add an authenticator app or a passkey as another way in.

People join by invitation. The first administrator claims the workspace, then invites everyone else with a role.

Your own settings live under You: your profile, how you sign in, your notifications and how Falcon looks.

## Features

### Sign in with an email code

Type your email address and enter the code Falcon sends you. There is no password to remember or leak.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Sign in from any browser with a one-time code
- Use the code alone, or add a second step if your company asks for one

In Falcon: Sign in · Second step

In the API: `getSignInOptions`, `startSignIn`, `completeSignIn`, `completeSecondStep`

### Authenticator apps and passkeys

Add an authenticator app or a passkey as another way to sign in, with backup codes for when you lose a device.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Add or remove an authenticator app or a passkey
- Create a new set of backup codes
- See where you are signed in and sign out of a session

In Falcon: Sign-in & security

In the API: `beginSecurityEnrollment`, `confirmSecurityEnrollment`, `removeSecurityMethod`, `regenerateBackupCodes`, `getUserSecurity`, `revokeUserSession`

### Sign in with Google or Microsoft

An administrator can let people sign in with their Google or Microsoft work account. Email codes stay available as a way back in.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Switch Google or Microsoft sign-in on for the company
- Link your Google or Microsoft account to your Falcon sign-in

In Falcon: Security · Sign-in & security

In the API: `getSignInOptions`, `startSignIn`, `completeSignIn`, `getCompanyMicrosoftSignIn`, `putCompanyMicrosoftSignIn`

### Set up the workspace

The administrator named when the workspace was created signs in once to claim it. Everyone else joins by invitation.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Claim the workspace as its first administrator
- Go straight on to invite your team or create the first agent

In Falcon: Set up the workspace

In the API: `getWorkspaceSetup`, `completeWorkspaceSetup`

### Join by invitation

People join a company in Falcon only when someone with the right role invites them. The invitation says which role they get.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Open an invitation link and join with your email
- Get the role the invitation names

In Falcon: Invitation

In the API: `getInvitation`, `acceptInvitation`

### Your profile, notifications and appearance

Keep your name and photo current, choose which notices you get, and pick light or dark.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Edit your name and photo
- Choose which notifications you get
- Pick light, dark or match your device

In Falcon: Profile · Notifications · Appearance

In the API: `getUserProfile`, `updateUserProfile`, `getUserPreferences`, `updateUserPreferences`

### Verify your phone number

Add a mobile number to your profile and confirm it with a code sent by text.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Add and confirm a mobile number

In Falcon: Profile

In the API: `startPhoneVerification`, `confirmPhoneVerification`

### Phone codes for two-step

Use a code sent to your mobile by text, RCS or a call to finish signing in and to confirm it's you. Passkeys and authenticator apps stay the stronger choice.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Add a mobile number as your second step and choose text, RCS or a call
- Get a code to sign in or confirm a change, with what it's for shown on screen
- Let your company turn phone codes off, or require passkeys or authenticator apps for administrators

In Falcon: Sign-in & security · Security

In the API: `startPhoneTwoStep`, `confirmPhoneTwoStep`, `setPhoneTwoStepDelivery`, `sendSecondStepPhoneCode`

### Ask to join a company

Someone with a company email address can ask to join, and an administrator says yes or no.

> Coming next: Designed and planned, not built yet.

What you can do:

- Ask to join from the sign-in page
- Approve or turn down requests as an administrator

## Who can do what

Who may do each thing in this guide, and what an AI assistant connected as that person may do with it. Roles are set in Company settings › Roles.

| What | Who may | An assistant |
|---|---|---|
| Read sign-in choices | The browser holding the required sign-in, setup, enrollment or consent challenge; callbacks also require their bound state. | Not offered to assistants |
| Start signing in | The browser holding the required sign-in, setup, enrollment or consent challenge; callbacks also require their bound state. | Not offered to assistants |
| Finish signing in | The browser holding the required sign-in, setup, enrollment or consent challenge; callbacks also require their bound state. | Not offered to assistants |
| Verify a second sign-in step | The browser holding the required sign-in, setup, enrollment or consent challenge; callbacks also require their bound state. | Not offered to assistants |
| Start adding a security method | The browser holding the required sign-in, setup, enrollment or consent challenge; callbacks also require their bound state. | Not offered to assistants |
| Verify a new security method | The browser holding the required sign-in, setup, enrollment or consent challenge; callbacks also require their bound state. | Not offered to assistants |
| Remove a security method | People who can edit it | Cannot be undone · needs “change settings” · waits for your approval |
| Replace backup codes | People who can edit it | Cannot be undone · needs “change settings” · waits for your approval |
| Reads your own sign-in methods and sessions | Anyone who can see it | Reads · needs “look” |
| Signs you out of one session, or everywhere | People who can edit it | Cannot be undone · needs “change settings” · waits for your approval |
| Read Microsoft sign-in settings | Company administrators | Reads · needs “look” |
| Save Microsoft sign-in settings | Company administrators | Changes · needs “change settings” · waits for your approval |
| Read workspace setup | The browser holding the required sign-in, setup, enrollment or consent challenge; callbacks also require their bound state. | Not offered to assistants |
| Finish workspace setup | The browser holding the required sign-in, setup, enrollment or consent challenge; callbacks also require their bound state. | Not offered to assistants |
| Read a membership invitation | The browser holding the required sign-in, setup, enrollment or consent challenge; callbacks also require their bound state. | Not offered to assistants |
| Accept a membership invitation | The browser holding the required sign-in, setup, enrollment or consent challenge; callbacks also require their bound state. | Not offered to assistants |
| Reads your own profile | Anyone who can see it | Reads · needs “look” |
| Changes your own profile | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Reads your own preferences | Anyone who can see it | Reads · needs “look” |
| Changes your own preferences | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Send a phone verification code | The browser holding the required sign-in, setup, enrollment or consent challenge; callbacks also require their bound state. | Not offered to assistants |
| Verify a phone number | The browser holding the required sign-in, setup, enrollment or consent challenge; callbacks also require their bound state. | Not offered to assistants |
| Send a code to add a phone for two-step | The browser holding the required sign-in, setup, enrollment or consent challenge; callbacks also require their bound state. | Not offered to assistants |
| Confirm the code and turn on phone two-step | The browser holding the required sign-in, setup, enrollment or consent challenge; callbacks also require their bound state. | Not offered to assistants |
| Choose how phone codes are sent | The browser holding the required sign-in, setup, enrollment or consent challenge; callbacks also require their bound state. | Not offered to assistants |
| Send a phone code for a second step | The browser holding the required sign-in, setup, enrollment or consent challenge; callbacks also require their bound state. | Not offered to assistants |

## Related

- Next: [Agents](https://falcon.com/docs/agents.md)
- [Glossary](https://falcon.com/docs/glossary.md)

---

Source: https://falcon.com/docs/agents

# Agents

Build an agent in plain words, give it skills, knowledge and tools, test it, then make a version Live.

An agent is a named assistant that talks to your customers or does work for your team. You shape it in plain words; there is no code to write.

Every agent has a draft and saved versions. Several people can edit the draft within their roles. Customers only ever meet the version you made Live.

What an agent can do comes from the parts you give it: instructions, skills, knowledge, tools, memory, voice and channels. Each has its own page inside the agent.

## Features

### The Agents page

See every agent you can open, grouped by category, with whether each is a draft, Live or waiting for approval.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Search and filter agents
- See the company's agents and your own side by side
- Start a new agent from a category

In Falcon: Agents

In the API: `listAgents`, `getAgent`, `getAgentOverview`

### Start a new agent

Start from a blank agent or a starter for a common job, then change anything you like.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Pick a starter or start blank
- Name the agent and choose its category and icon
- Base it on another agent so it inherits that agent's setup

In Falcon: New agent

In the API: `createAgent`, `createAgentFromTemplate`, `listAgentBases`

### Instructions

Tell the agent who it is, what it is for and how to behave, in plain words, with a preview of what it will be told.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Write and edit the agent's instructions
- Preview the instructions the agent receives
- Set how it should write on each channel, such as shorter on texts

In Falcon: Instructions · How it sounds on each channel

In the API: `getAgentInstructions`, `updateAgentInstructions`, `getAgentChannelStyles`, `updateAgentChannelStyles`

### Skills

A skill is a reusable piece of know-how, with a note on when to use it. Write one once and share it across agents.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Write a company skill and publish it
- Add skills to an agent
- Say when the agent should use each skill

In Falcon: Skills · Skill

In the API: `listCompanySkills`, `createCompanySkill`, `updateCompanySkill`, `publishCompanySkill`, `getAgentSkills`, `updateAgentSkills`

### Knowledge

Give agents your help articles, policies and documents to answer from. Falcon keeps its copy up to date.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Upload files, or point at a website, Google Drive, SharePoint, GitHub or Amazon S3
- Choose which knowledge bases each agent searches
- Try a search to see what the agent would find

In Falcon: Knowledge · New knowledge base · Knowledge base · Knowledge searches

In the API: `listKnowledgeBase`, `createKnowledgeBase`, `createKnowledgeSource`, `startKnowledgeSync`, `testKnowledgeSearch`, `getAgentKnowledgeSearches`, `updateAgentKnowledgeSearches`

### Tools

Tools let an agent do things, such as look something up, save a file or send a message. You choose which tools each agent may use.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Add built-in tools to an agent
- Set a tool's defaults once for the company
- Require a person to approve a tool before it is used

In Falcon: Tools · New tool · Tool

In the API: `listToolCatalog`, `listCompanyTool`, `getAgentTools`, `updateAgentTools`, `updateAgentToolSettings`, `approveAgentTool`

### Your own endpoints and MCP servers as tools

Turn your own web endpoint or a remote MCP server into a tool your agents can call.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Describe an endpoint and test it
- Connect a remote MCP server and pick its tools
- Sign requests so your endpoint can trust them

In Falcon: New tool · Tool · Tool sign-in

In the API: `createCompanyTool`, `testCompanyTool`, `createRemoteMcpConnection`, `selectConnectorTools`, `putToolWebhookSigning`

### Messaging tools

Tools for sending texts, WhatsApp, Slack, Telegram and email, with company rules for each.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Set who an email comes from and where replies go
- Choose what happens when a reply could be for more than one conversation
- Keep every message answerable, so replies find the right agent

In Falcon: Messaging

In the API: `getTextMessagingSettings`, `putTextMessagingSettings`, `getEmailMessagingSettings`, `putEmailMessagingSettings`, `getSlackMessagingSettings`, `getWhatsappMessagingSettings`, `getTelegramMessagingSettings`

### Memory

Choose what an agent remembers about the people it talks to, and send what it learns to your own systems.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Choose the kinds of memory an agent keeps
- Send new memories to your own endpoint

In Falcon: Memory · Memory type

In the API: `getAgentMemory`, `updateAgentMemory`, `listMemoryDeliveries`, `createMemoryWebhook`

### Variables

Give an agent named values, such as an opening time or a booking link, that its instructions can use.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Add, change and remove an agent's variables

In Falcon: Variables

In the API: `getAgentVariables`, `updateAgentVariables`

### Choose the model

Pick which AI model an agent uses from the models your company allows.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Choose the model for each agent
- See the models the company has approved

In Falcon: Settings

In the API: `listModelChoices`, `getAgentSettings`, `updateAgentSettings`

### Voice

Choose whether an agent is off on calls, only listens and takes notes, or listens and talks.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Set the voice mode: off, listen only, or listen and talk
- Offer a choice of voices, and let each person pick the one they hear

In Falcon: Voice

In the API: `getAgentVoiceSettings`, `putAgentVoiceSettings`

### Channels

Choose where people can reach an agent: a chat widget on your website, a link, texts and RCS, WhatsApp, Slack, Telegram or email.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Switch each channel on or off
- Style the website chat widget to match your brand
- Share a public link to talk to the agent

In Falcon: Channels · Channel

In the API: `getAgentChannels`, `putAgentChannels`, `getAgentPublicEntry`, `putAgentPublicEntry`, `getAgentWidgetSkin`, `putAgentWidgetSkin`

### Ask an admin

When an agent is unsure, it asks a person instead of guessing, and tells them only what they need to answer.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Choose who the agent asks
- See exactly what the admin is told

In Falcon: Escalation

In the API: `getAgendaSettings`, `putAgendaSettings`, `agenda_list`, `agenda_update`

### Who an agent's messages come from

Choose the identity an agent speaks as by default, such as a person on your team or one of your places.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Pick the agent's default identity
- See how the identity's details appear in messages

In Falcon: Identity

### Test an agent

Talk to the draft before anyone else does, look inside each reply, and save good conversations as test cases.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Chat with the draft
- Inspect the steps behind a reply
- Save a conversation as a test case and run it again later

In Falcon: Test

In the API: `createAgentTestRun`, `inspectAgentTestRun`, `saveAgentTestCase`, `listAgentTestCases`, `listAgentTestRuns`, `getAgentTestChoices`

### Make a version Live

Nothing reaches customers until someone with permission makes a version Live. Falcon checks the agent is ready first.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- See what is missing before publishing
- Make a version Live, or send it for approval

In Falcon: Publish

In the API: `getAgentPublish`, `publishAgent`

### Versions

Every published version is kept, so you can see what changed between two of them.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- List an agent's versions
- Compare two versions side by side

In Falcon: Publish

In the API: `listAgentVersions`, `compareAgentVersions`

### Access and collaborators

Decide who can see, edit or publish each agent, and invite collaborators to one agent only.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Share an agent with people or roles
- Invite a collaborator to one agent
- See who can do what, and why

In Falcon: Access

In the API: `getAgentAccess`, `updateAgentAccess`, `getEffectiveAgentAccess`, `inviteAgentCollaborator`, `requestAgentInvitation`, `leaveSharedAgent`

### Built on this

Let other agents build on this one, so a change here reaches them too, and see what a change would affect.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Allow or stop other agents building on this one
- See which agents would be affected by a change

In Falcon: Built on this

In the API: `getAgentBuiltOn`, `updateAgentBuiltOn`, `getBuiltOnImpact`, `getAgentInheritance`

### Activity

See what an agent has been doing. The words of a conversation stay hidden until someone with permission chooses to show them.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Browse recent conversations and runs
- Show a conversation's words, which is recorded
- Replay the steps an agent took

In Falcon: Activity

In the API: `revealConversationWords`, `replayConversationSteps`

### Pause, copy, hand over or archive

Pause an agent, copy it, hand it to someone else, archive it or take its content with you.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Pause and restart an agent
- Duplicate an agent
- Transfer an agent to another person
- Archive, restore or delete an agent
- Export an agent's content

In Falcon: Settings

In the API: `pauseAgent`, `duplicateAgent`, `transferAgent`, `archiveAgent`, `restoreAgent`, `deleteAgent`, `exportAgentContent`

### Session rules

Rules for how long a conversation lasts and when a new one begins.

> Coming next: Designed and planned, not built yet.

What you can do:

- Set when a conversation ends
- Decide what the agent carries into the next one

### Script tools

Write a short script as a tool, with its own settings and secrets, and let agents call it.

> Coming next: Designed and planned, not built yet.

What you can do:

- Write and test a script tool
- Keep company variables and secrets for tools in one place

### Rate replies and improve them

Mark a reply good or bad and ask Falcon to suggest a change to the instructions.

> Coming next: Designed and planned, not built yet.

What you can do:

- Rate a reply
- Get a suggested improvement to review

### Push notifications

Reach people through notifications on their phone, as another channel.

> Coming next: Designed and planned, not built yet.

What you can do:

- Send a notification that opens the conversation

## Who can do what

Who may do each thing in this guide, and what an AI assistant connected as that person may do with it. Roles are set in Company settings › Roles.

| What | Who may | An assistant |
|---|---|---|
| Lists the agents you can see, with the ones you can open and the ones you can only see the name of | Anyone who can see it | Reads · needs “look” |
| Read an agent | Anyone who can see it | Reads · needs “look” |
| Read an agent overview | Anyone who can see it | Reads · needs “look” |
| Creates a new agent | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Create agent from template | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Lists the agents a new agent can be built on | Anyone who can see it | Reads · needs “look” |
| Read agent instructions | Anyone who can see it | Reads · needs “look” |
| Changes an agent's instructions | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Read agent writing styles by channel | Anyone who can see it | Reads · needs “look” |
| Changes how an agent writes on each channel | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Lists the company's skills | Anyone who can see it | Reads · needs “look” |
| Creates a company skill | People who manage skills | Changes · needs “change settings” · waits for your approval |
| Changes a company skill's draft | People who manage skills | Changes · needs “change settings” · waits for your approval |
| Publishes a skill's draft as a new version | People who can publish it | Reaches outside · needs “change settings” · waits for your approval |
| Read skills bound to an agent | Anyone who can see it | Reads · needs “look” |
| Changes the skills an agent uses and when it uses them | People who manage skills | Changes · needs “change settings” · waits for your approval |
| List knowledge bases | Anyone who can see it | Reads · needs “look” |
| Creates a knowledge base | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Adds a source to a knowledge base | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Starts syncing a knowledge base from its sources | People who can edit it | Reaches outside · needs “run and test” · waits for your approval |
| Test knowledge search | People who can edit it | Reaches outside · needs “run and test” · waits for your approval |
| Read agent knowledge searches | Anyone who can see it | Reads · needs “look” |
| Update agent knowledge searches | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Lists the tools and knowledge searches you can add to an agent | Anyone who can see it | Reads · needs “look” |
| List company tools | Anyone who can see it | Reads · needs “look” |
| Read tools assigned to an agent | Anyone who can see it | Reads · needs “look” |
| Changes which tools and knowledge searches an agent has | People who manage tools | Changes · needs “change settings” · waits for your approval |
| Update agent tool settings | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Approves or rejects a tool that is waiting for review on an agent | People who can approve it | Changes · needs “change settings” · waits for your approval |
| Defines a new company tool | People who manage tools | Changes · needs “change settings” · waits for your approval |
| Runs a test call of a company tool. Repeat the same call with the same idempotency key to read its result | People who manage tools | Reaches outside · needs “run and test” · waits for your approval |
| Adds a connection to another MCP server | People who manage tools | Changes · needs “change settings” · waits for your approval |
| Select connector tools | People who manage tools | Changes · needs “change settings” · waits for your approval |
| Put tool webhook signing | People who manage tools | Changes · needs “change settings” · waits for your approval |
| Read text shared-sender reply settings | Anyone who can see it | Reads · needs “look” |
| Set text shared-sender reply settings | People who manage tools | Changes · needs “change settings” |
| Read email shared-sender reply settings | Anyone who can see it | Reads · needs “look” |
| Set email shared-sender reply settings | People who manage tools | Changes · needs “change settings” |
| Read Slack shared-sender reply settings | Anyone who can see it | Reads · needs “look” |
| Read WhatsApp shared-sender reply settings | Anyone who can see it | Reads · needs “look” |
| Read Telegram shared-sender reply settings | Anyone who can see it | Reads · needs “look” |
| Read agent memory settings | Anyone who can see it | Reads · needs “look” |
| Changes what an agent remembers and how | People who can edit it | Changes · needs “change settings” · waits for your approval |
| List memory deliveries | Anyone who can see it | Reads · needs “look” |
| Create memory webhook | People who can edit it | Reaches outside · needs “change settings” · waits for your approval |
| Read agent variable definitions | Anyone who can see it | Reads · needs “look” |
| Changes an agent's variables | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Lists the models you can choose for an agent | Anyone who can see it | Reads · needs “look” |
| Read agent settings | Anyone who can see it | Reads · needs “look” |
| Changes an agent's settings, voice and model | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Read agent voice settings | Anyone who can see it | Reads · needs “look” |
| Put agent voice settings | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Read agent email and phone channels | Anyone who can see it | Reads · needs “look” |
| Put agent channels | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Read agent public entry settings | Anyone who can see it | Reads · needs “look” |
| Put agent public entry | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Read an agent's chat window skin | Anyone who can see it | Reads · needs “look” |
| Change an agent's chat window skin | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Read agenda tool settings | Anyone who can see it | Reads · needs “look” |
| Changes the company's settings for the agenda tool | People who manage tools | Changes · needs “change settings” · waits for your approval |
| Agenda list | People who manage tools | Reads · needs “look” |
| Agenda update | People who manage tools | Changes · needs “change settings” · waits for your approval |
| Create agent test run | People who can edit it | Reaches outside · needs “run and test” · waits for your approval |
| Inspect agent test run | Anyone who can see it | Reads · needs “look” |
| Save agent test case | People who can edit it | Changes · needs “change settings” · waits for your approval |
| List agent test cases | Anyone who can see it | Reads · needs “look” |
| List agent test runs | Anyone who can see it | Reads · needs “look” |
| Read agent test choices | Anyone who can see it | Reads · needs “look” |
| Read agent publication readiness | Anyone who can see it | Reads · needs “look” |
| Publishes the agent's current draft as a new version | People who can publish it | Reaches outside · needs “change settings” · waits for your approval |
| List saved agent versions | Anyone who can see it | Reads · needs “look” |
| Compares two versions of an agent and shows what changed | Anyone who can see it | Reads · needs “look” |
| Read agent sharing settings | Anyone who can see it | Reads · needs “look” |
| Changes who can see and edit an agent | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Read a person's effective agent access | Anyone who can see it | Reads · needs “look” |
| Invites a person to work on an agent | People who can edit it | Reaches outside · needs “change settings” · waits for your approval |
| Asks for access to an agent you can see but not open | People who can edit it | Reaches outside · needs “change settings” · waits for your approval |
| Removes your own access to an agent someone shared with you | People who can edit it | Cannot be undone · needs “change settings” · waits for your approval |
| Read an agent base selection | Anyone who can see it | Reads · needs “look” |
| Changes which agent this agent is built on | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Read the impact of an agent base | Anyone who can see it | Reads · needs “look” |
| Read agent instruction inheritance | Anyone who can see it | Reads · needs “look” |
| Reveal conversation words | People who can reveal secrets | Changes · needs “change settings” · waits for your approval |
| Replay conversation steps | People who can edit it | Reaches outside · needs “run and test” · waits for your approval |
| Pause agent | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Duplicate agent | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Transfer agent | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Archive agent | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Restore agent | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Delete an agent | People who can delete it | Cannot be undone · needs “change settings” · waits for your approval |
| Export agent content | People who can edit it | Changes · needs “change settings” · waits for your approval |

## Related

- Next: [Identities](https://falcon.com/docs/identities.md)
- [Glossary](https://falcon.com/docs/glossary.md)

---

Source: https://falcon.com/docs/identities

# Identities

The people and places your agents speak as, each with its own details, hours and ways to be reached.

An identity is a person or a place that an agent acts for, such as Dana in sales or the Downtown store. One agent can speak as many identities.

Identities are not Falcon users. They never sign in. They are records the company keeps, with the details an agent needs to speak as them.

The company decides which details every identity has, such as a name, a communication style or a background. An agent is told those details when it writes as that identity.

## Features

### Company identities

Keep the company's people and places in one list, each with its own page and a short code you can use to refer to it.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Add a person or a place
- Give each identity a code that stays stable if you rename it
- Explain on the page how identities differ from users

In Falcon: Identities · New identity · Identity

### Your own identities

Keep identities of your own, for agents you build for yourself.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Add and edit identities only you use

In Falcon: Yours · New identity of your own · Identity

### Identity fields

Decide once which details every identity has. Name, company, communication style, background and voice are built in; you can hide them or add your own.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Add a field and say what the agent is told about it
- Hide a built-in field you don't need
- Mark a field required

In Falcon: Fields

### Checks on identity details

Falcon checks what is typed into an identity's fields, such as whether a name looks like a name, before it is saved.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Choose which fields are checked
- See why a value was refused or flagged

In Falcon: Fields · Checks

In the API: `getValidationChecks`, `tryValidationCheck`

### Working hours

Give a person or place its hours, so workflows know when it is open.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Set opening or working hours for each day
- Use the hours in a workflow's hours check

In Falcon: Identity · New identity

### Where an identity can be reached

Each identity an agent acts as can have its own email address, phone number, voice address, link and website chat widget.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- See each address and copy the widget code
- Send a test to check an address works

In Falcon: Identity

In the API: `createCompanyIdentityBindingTest`, `getCompanyIdentityBindingTest`, `createPersonalIdentityBindingTest`, `getPersonalIdentityBindingTest`

### RCS for each identity

Messages can come from the company's shared RCS sender, or an identity can have its own.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Choose the shared sender or an identity's own
- Preview the registration details before applying

In Falcon: Identity

In the API: `getCompanyIdentityBindingRcs`, `updateCompanyIdentityBindingRcs`, `previewCompanyIdentityBindingRcsRegistration`

### An identity's own mailbox and calendar

Connect an identity's own mailbox and calendar, so an agent can send as them and check when they are free.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Connect a Gmail mailbox for an identity
- Connect a calendar and choose what the agent may see

In Falcon: Identity

In the API: `getCompanyIdentityBindingCalendar`, `updateCompanyIdentityBindingCalendar`, `getPersonalIdentityBindingCalendar`, `updatePersonalIdentityBindingCalendar`

### Images of an identity

Add photos of a person or place, with a note on when an agent should use each one.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Upload images with captions
- Say when each image should be used

In Falcon: Identity

### Import identities

Bring in many identities at once from a file. Falcon shows every row before anything is saved.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Upload a CSV or JSON file
- Match its columns to identity fields, or ignore a column
- Preview every row, then get a receipt

In Falcon: Import identities

In the API: `previewCompanyIdentityImport`, `commitCompanyIdentityImport`

### How the company's email looks

Set the layout and logo placement for email your agents send, once for the whole company.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Choose a layout and where the logo goes
- Preview an email before it is sent

In Falcon: Email

### Outlook mailboxes and calendars

Connect an identity's Outlook mailbox and Microsoft calendar, as with Gmail.

> Coming next: Designed and planned, not built yet.

What you can do:

- Send as an identity from Outlook
- Check a Microsoft calendar for free time

### A customer's contact points

See every way a customer has reached you, which agent and identity answered, and when.

> Coming next: Designed and planned, not built yet.

What you can do:

- See a person's numbers and addresses in one place
- See which conversation each belongs to

### Pictures drawn in an identity's likeness

Let an agent draw a picture of an identity from its reference images, labelled as made by AI.

> Coming next: Designed and planned, not built yet.

What you can do:

- Choose a primary image and style note
- Label every drawn image as made by AI

## Who can do what

Who may do each thing in this guide, and what an AI assistant connected as that person may do with it. Roles are set in Company settings › Roles.

| What | Who may | An assistant |
|---|---|---|
| Read company validation checks | Company administrators | Reads · needs “look” |
| Try a validation check | Company administrators | Changes · needs “change settings” · waits for your approval |
| Request a company binding address test | Company administrators | Changes · needs “run and test” · waits for your approval |
| Read a company binding address test | Company administrators | Reads · needs “look” |
| Request a personal binding address test | People who can edit it | Changes · needs “run and test” · waits for your approval |
| Read a personal binding address test | Anyone who can see it | Reads · needs “look” |
| Read an identity binding's RCS sender | Company administrators | Reads · needs “look” |
| Choose an identity binding's RCS sender | Company administrators | Changes · needs “change settings” · waits for your approval |
| Preview an identity RCS registration | Company administrators | Changes · needs “change settings” |
| Read company binding calendar status | Company administrators | Reads · needs “look” |
| Choose company binding calendars | Company administrators | Changes · needs “change settings” |
| Read personal binding calendar status | Anyone who can see it | Reads · needs “look” |
| Choose personal binding calendars | People who can edit it | Changes · needs “change settings” |
| Preview a company identity import | Company administrators | Changes · needs “change settings” |
| Commit a company identity import | Company administrators | Changes · needs “change settings” · waits for your approval |

## Related

- Next: [Workflows](https://falcon.com/docs/workflows.md)
- [Glossary](https://falcon.com/docs/glossary.md)

---

Source: https://falcon.com/docs/workflows

# Workflows

Chain agents and steps together, start them from events, and decide what happens when someone replies.

A workflow is a series of steps that runs when something happens, such as a new email, a form, a call or a time of day. Each step is an agent or a check.

Falcon shows each workflow as one sentence you can read, and as steps you can open and change.

Like agents, workflows have drafts and a Live version. You can try a workflow before it runs for real.

## Features

### Workflows and how they work

See the company's workflows, and read a short guide to how a workflow runs.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Browse workflows and their state
- Read How workflows work
- Describe a new workflow in your own words and let Falcon draft it

In Falcon: Workflows · How workflows work

In the API: `listPipelines`

### Edit a workflow

Add, move and remove steps on a canvas, with the chosen step's settings beside it.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Add an agent step and choose its tools
- Reorder or remove steps
- Try the workflow before it is Live

In Falcon: Edit

In the API: `getPipelineVersion`, `testPipeline`, `updateStepToolSettings`, `getStepToolSettings`

### How a workflow starts

Start a workflow on a schedule, when an app sends an event, when a conversation arrives, or when another system calls it.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Pick a start from connected apps first
- Set the start's details inline, with examples
- See an example of the event that starts it

In Falcon: Edit

In the API: `listPipelineTriggers`, `createPipelineTrigger`, `getPipelineEventExample`

### Side chains and retries

Hang extra steps off a step for when someone is interested, replies STOP, or the step fails.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Add a side chain to any step
- Retry a failed step, or run a side chain instead

In Falcon: Edit

### Hours check

A step that checks whether a person or place is open now, and sends the workflow one way or the other.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Check an identity's working hours
- Send out-of-hours work down its own path

In Falcon: Edit

### Front desk

A first step that greets the person, looks them up in past conversations, and hands them to the right agent, such as sales or service.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Name the routes and what each is for
- Match the person to past conversations first
- Send each route to its own agent

In Falcon: Edit

### Engagements

Say what counts as someone engaging, such as replying or booking, so each can start a side chain.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Define what the company counts as engaging
- Use an engagement to start a side chain

In Falcon: Engagements

### Waiting for you

Drafts that need a person before they are sent, from every workflow, in one place.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Send a draft as it is
- Edit a draft, then send it
- Decide not to send it

In Falcon: Waiting for you

In the API: `adoptGmailReplyDraftTemplate`, `installGmailReplyDraftTemplate`

### Request mapping

Give any outside system its own address to call, and map what it sends into a workflow or an agent.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Create an address for each system and rotate it when needed
- Map fields from the request into the workflow
- Start from ready-made settings for Twilio, Zoom or RingCentral recordings
- Check Stripe, Shopify, GitHub and Slack requests against the signature each one sends
- Check signatures or a shared key before anything runs
- Map each value from a body field, query parameter, header or fixed value into a start, identity or customer field, or an outgoing header
- See what the target receives, the steps that get it first, and what your system gets back
- Flag maps for a look when a published workflow changes its start fields or result

In the API: `listIncomingRequestMaps`, `listIncomingRequestTargets`, `createIncomingRequestMap`, `updateIncomingRequestMap`, `rotateIncomingRequestMap`, `receiveIncomingRequest`

### Runs

See each time a workflow ran, and step in when one needs it.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Retry, cancel or reschedule a run
- Run a scheduled run now

In Falcon: Runs

In the API: `retryPipelineRun`, `cancelPipelineRun`, `reschedulePipelineRun`, `runnowPipelineRun`

### Publish and roll back

Make a workflow Live, go back to an earlier version, or archive it.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Make a version Live
- Roll back to the previous version
- Archive a workflow you no longer use

In Falcon: Publish

In the API: `getPipelineVersion`, `rollbackPipeline`, `archivePipeline`

### Webhooks out

Tell your own systems when a workflow does something, by sending them a signed message.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.
>
> Available since 29 September 2026.

What you can do:

- Add an address to notify
- Choose which events it receives

In Falcon: Webhooks out

In the API: `listPipelineWebhooks`, `createPipelineWebhook`, `updatePipelineWebhook`

### When a reply could be for more than one

If two agents or identities wrote to someone and they reply, choose whether the reply goes to the latest, to both, or is read to decide.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Set the rule for each channel
- Override it on a workflow step

In the API: `getTextMessagingSettings`, `putTextMessagingSettings`

### Taking turns

Share new conversations between identities in turn, such as sales reps who are available, and keep returning people with the same one.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Rotate new leads between identities in turn, to the least busy, or by weight
- Only give turns to identities that are active, inside their working hours and free on their calendar
- Set a daily limit per identity, and choose who answers or wait when nobody is available
- Keep a returning customer with their rep on any channel, and switch them when the rep leaves
- Move a person to someone else by hand
- See conversations per identity and why each one got them

In Falcon: Identity

In the API: `getIdentityTurns`, `findIdentityTurn`, `reassignIdentityTurn`

### Acting as on a step

Pick which identity an agent acts as on each step, so the same agent can appear three times as three people.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Choose an identity for each agent step, or leave it to the workflow
- Let a step take turns among identities, keeping returning people with the same one
- Use the step's identity for its tools, what it remembers and the messages it sends
- Commit to GitHub as the identity, with the requester credited

In Falcon: Edit

### Topics and keywords on messages

Label outgoing messages with a topic, so opting out of marketing never stops order updates, and route replies by keyword.

> Coming next: Designed and planned, not built yet.

What you can do:

- Keep a list of topics for the company
- Let people opt out of one topic only
- Route a reply by keyword

In the API: `getCompanyMessageTopics`, `updateCompanyMessageTopics`

### Evaluate steps

A step that scores or checks the work of earlier steps, and workflows that only evaluate.

> Coming next: Designed and planned, not built yet.

What you can do:

- Check a plan is complete before it is sent
- Run an evaluation on its own

### Sequences that repeat

Repeat a set of steps, such as a follow-up every few days, until someone replies or a limit is reached.

> Coming next: Designed and planned, not built yet.

What you can do:

- Repeat steps on a schedule
- Stop when the person replies

### Nudges and review rounds

Remind people who haven't answered, and pass work through more than one review before it goes out.

> Coming next: Designed and planned, not built yet.

What you can do:

- Send a reminder after a set time
- Add review rounds

## Who can do what

Who may do each thing in this guide, and what an AI assistant connected as that person may do with it. Roles are set in Company settings › Roles.

| What | Who may | An assistant |
|---|---|---|
| Lists the company's workflows | Anyone who can see it | Reads · needs “look” |
| Get workflow version | Anyone who can see it | Reads · needs “look” |
| Test workflow | People who can edit it | Reaches outside · needs “run and test” · waits for your approval |
| Update step tool settings | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Read workflow step tool settings | Anyone who can see it | Reads · needs “look” |
| Lists what starts a workflow | Anyone who can see it | Reads · needs “look” |
| Adds something that starts a workflow | People who can edit it | Reaches outside · needs “change settings” · waits for your approval |
| Get workflow event example | Anyone who can see it | Reads · needs “look” |
| Adopt a Gmail reply draft agent | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Install the Gmail reply draft workflow | People who can edit it | Changes · needs “change settings” · waits for your approval |
| List incoming request maps | Anyone who can see it | Reads · needs “look” |
| List what request map targets receive and give back | Anyone who can see it | Reads · needs “look” |
| Create an incoming request map | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Update an incoming request map | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Rotate an incoming request address | People who can rotate secrets | Changes · needs “change settings” · waits for your approval |
| Receive a mapped incoming request | The configured sender verification; no Studio session is accepted. | Not offered to assistants |
| Retry workflow run | People who can edit it | Reaches outside · needs “run and test” · waits for your approval |
| Cancel workflow run | People who can edit it | Cannot be undone · needs “run and test” · waits for your approval |
| Reschedule workflow run | People who can edit it | Changes · needs “run and test” · waits for your approval |
| Runnow workflow run | People who can edit it | Reaches outside · needs “run and test” · waits for your approval |
| Restores an older workflow version as the draft. To change Live, make an older version live instead | People who can make it Live | Reaches outside · needs “change settings” · waits for your approval |
| Archive a workflow | People who can edit it | Cannot be undone · needs “change settings” · waits for your approval |
| Lists a workflow's webhooks | Anyone who can see it | Reads · needs “look” |
| Adds a webhook to a workflow | People who can edit it | Reaches outside · needs “change settings” · waits for your approval |
| Changes a workflow webhook | People who can edit it | Reaches outside · needs “change settings” · waits for your approval |
| Read text shared-sender reply settings | Anyone who can see it | Reads · needs “look” |
| Set text shared-sender reply settings | People who manage tools | Changes · needs “change settings” |
| Report on taking turns | Anyone who can see it | Reads · needs “look” |
| Find who a person is with | People who can edit it | Reads · needs “look” |
| Move a person to another identity | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Read company message topics | Company administrators | Reads · needs “look” |
| Update company message topics | Company administrators | Changes · needs “change settings” · waits for your approval |

## Related

- Next: [Connections](https://falcon.com/docs/connections.md)
- [Glossary](https://falcon.com/docs/glossary.md)

---

Source: https://falcon.com/docs/connections

# Connections

Connect the apps, mailboxes, numbers and storage your agents and workflows use.

A connection is how Falcon reaches something outside it: an app such as a CRM, a mailbox, a phone provider or a storage bucket.

You connect a service once for the company. Then you choose which tools agents may use and which events can start workflows.

Some connections belong to one person, such as your own mailbox. Those appear under Your connections, and only you can use them.

Other systems can also call Falcon. See request mapping in the Workflows guide.

## Features

### App connections

Connect business apps such as a CRM, payments or email with a browser sign-in, using Falcon's sign-in or your own.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Add an app and sign in to it
- Choose which of its tools agents may use
- Choose which of its events can start workflows

In Falcon: Connections · Connect a service · Finishing connecting

In the API: `listConnectorPresets`, `getConnectorPreset`, `connectConnectorPreset`, `selectConnectorTools`, `serviceListConnectedApps`, `getCompanyAppSettings`, `updateCompanyAppSettings`

### Your connections

Connections that belong to you, such as your own mailbox, and what may use them.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- See the apps connected in your name
- Disconnect any of them

In Falcon: Your connections

In the API: `listConnectedApps`, `disconnectConnectedApp`

### Email sending

Send your agents' email from your own domain, through SendGrid, or through any mailbox that offers SMTP and IMAP.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Connect SendGrid and send a test
- Connect another mailbox by SMTP, with IMAP to read replies
- See whether each sender is ready

In Falcon: Email

In the API: `createSendGridConnection`, `testSendGridConnection`, `listSendGridConnection`, `createOtherEmailConnection`, `listOtherEmailConnections`, `disconnectOtherEmailConnection`, `listSendingSenders`

### Twilio

Connect your Twilio account to send and receive texts with your own numbers.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Connect a Twilio account and check it works
- See the senders it offers

In Falcon: Twilio

In the API: `createTwilioConnection`, `putTwilioConnection`, `checkTwilioConnection`, `listTwilioConnections`

### Phone numbers and AI voicemail

Give an identity its own phone number for incoming calls, and let calls outside hours go to an AI voicemail that takes a message.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Link a number to an identity
- Set a number to answer as an AI voicemail
- Detach or release a number

In Falcon: Numbers

### Storage and files

Connect an Amazon S3 bucket so agents can save, read and list files, and share files by link.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Connect a bucket and check access
- Create file stores with their own rules
- Let agents save and read files

In Falcon: Storage

In the API: `createStorageConnection`, `checkStorageConnection`, `createFileStore`, `updateFileStore`, `createFileLink`

### Event sources

Let a system send events to Falcon, so a workflow can start when something happens there.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Add an event source and get its address
- See an example event

In Falcon: Event sources · An event source

In the API: `getCompanyWebhookEndpoint`, `updateCompanyWebhookEndpoint`, `getAppCloudEventExample`

### Domains and links

Set which domains the company's links and addresses use.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- See and set the company's domains

In Falcon: Domains & links

In the API: `getCompanyDomainPolicy`, `updateCompanyDomainPolicy`

### GitHub repositories

Install the GitHub app for your organisation and choose which repositories agents may work in.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Install on a GitHub organisation
- Pick repositories for agents to use

In Falcon: GitHub repositories

In the API: `createCompanyGithubInstallation`, `actOnCompanyGithubInstallation`

### Your own REST API

Describe your own REST API once and turn its endpoints into tools.

> Coming next: Designed and planned, not built yet.

What you can do:

- Import an API description
- Choose which endpoints become tools

### Message keys

Keys that let your own systems send messages through your agents' channels.

> Coming next: Designed and planned, not built yet.

What you can do:

- Create a key and limit what it may send

## Who can do what

Who may do each thing in this guide, and what an AI assistant connected as that person may do with it. Roles are set in Company settings › Roles.

| What | Who may | An assistant |
|---|---|---|
| List connection presets | Anyone who can see it | Reads · needs “look” |
| Read a connection preset | Anyone who can see it | Reads · needs “look” |
| Connect connector preset | People who can edit it | Reaches outside · needs “change settings” · waits for your approval |
| Select connector tools | People who manage tools | Changes · needs “change settings” · waits for your approval |
| List connected apps | Anyone who can see it | Reads · needs “look” |
| Read company app sign-in settings | Anyone who can see it | Reads · needs “look” |
| Change company app sign-in settings | People who can edit it | Changes · needs “change settings” · waits for your approval |
| List connected apps | Anyone who can see it | Reads · needs “look” |
| Disconnect connected app | People who can edit it | Cannot be undone · needs “change settings” · waits for your approval |
| Create send grid connection | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Test send grid connection | People who can edit it | Reaches outside · needs “run and test” · waits for your approval |
| List SendGrid connections | Anyone who can see it | Reads · needs “look” |
| Connect an Other email mailbox | People who can edit it | Reaches outside · needs “change settings” · waits for your approval |
| Read saved Other email mailboxes | Anyone who can see it | Reads · needs “look” |
| Disconnect an Other email mailbox | People who can delete it | Changes · needs “change settings” · waits for your approval |
| List available message senders | Anyone who can see it | Reads · needs “look” |
| Create twilio connection | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Put twilio connection | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Check twilio connection | People who can edit it | Reaches outside · needs “change settings” · waits for your approval |
| List twilio connections | Anyone who can see it | Reads · needs “look” |
| Adds a storage connection | Company administrators | Changes · needs “change settings” · waits for your approval |
| Checks that a storage connection works | Company administrators | Reaches outside · needs “change settings” · waits for your approval |
| Creates a file store | Company administrators | Changes · needs “change settings” · waits for your approval |
| Changes a file store | Company administrators | Changes · needs “change settings” · waits for your approval |
| Creates a link to a stored file | Company administrators | Changes · needs “change settings” · waits for your approval |
| Read the company webhook endpoint | Anyone who can see it | Reads · needs “look” |
| Update company webhook endpoint | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Read an app event example | Anyone who can see it | Reads · needs “look” |
| Read email domain policy | Company administrators | Reads · needs “look” |
| Change email domain policy | Company administrators | Changes · needs “change settings” · waits for your approval |
| Add a GitHub installation | Company administrators | Reaches outside · needs “change settings” · waits for your approval |
| Run a GitHub installation action | Company administrators | Cannot be undone · needs “change settings” · waits for your approval |

## Related

- Next: [Company settings](https://falcon.com/docs/company-settings.md)
- [Glossary](https://falcon.com/docs/glossary.md)

---

Source: https://falcon.com/docs/company-settings

# Company settings

People, roles, security, models and spend limits, billing, the audit log and your data.

Company settings are for administrators. They apply to everyone in the company.

Each page makes one kind of decision: who is here, what they may do, how they sign in, which models agents may use, and how much they may spend.

Every change is recorded in the audit log.

## Features

### General and brand

The company's name, logo, colours and images, previewed before you apply them.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Change the company name and logo
- Preview a new theme, apply it or go back
- Keep a library of company images

In Falcon: General · Images

In the API: `getCompanyTheme`, `changeCompanyThemeDraft`, `changeCompanyThemePreview`, `changeCompanyThemeApply`, `changeCompanyThemeRevert`, `uploadCompanyLogo`, `listCompanyImage`, `createCompanyImage`

### People

See everyone in the company, invite someone new, and make a person inactive when they leave.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Invite someone with a role
- Change a person's role
- Make a person inactive and sign them out

In Falcon: People · Invite someone · A person

In the API: `listCompanyPerson`, `searchCompanyPeople`, `createCompanyInvitation`, `updateCompanyPerson`, `signOutCompanyPerson`

### Invite many people at once

Upload a list of people with their roles and send all the invitations together.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Upload a CSV of up to 500 people
- Check every row before anything is sent
- See which invitations were sent and which failed

In Falcon: Invite someone

In the API: `invitePeople`

### Roles

A role is a set of things a person may do. Use the built-in roles or make your own.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- See what each role allows
- Create or change a role

In Falcon: Roles · A role

In the API: `listRoles`

### Security

Decide how people sign in, whether a second step is required, and how long sessions last.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Choose the ways people may sign in
- Require a second step
- Set session length

In Falcon: Security

In the API: `actOnCompanySecurity`

### Single sign-on

Let people sign in through the company's own Microsoft tenant or identity provider.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.
>
> Plans: only on the Enterprise plan.

What you can do:

- Approve Falcon in your Microsoft tenant, or use your own app
- Set up single sign-on for your domain

In Falcon: Single sign-on

In the API: `actOnCompanySso`, `getCompanyMicrosoftSignIn`, `putCompanyMicrosoftSignIn`, `completeCompanyMicrosoftConsent`

### Checks

One place for the checks Falcon runs on what people type, such as identity details, with a way to try each one.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Turn checks on or off
- Adjust how strict a check is
- Try a check on your own example

In Falcon: Checks

In the API: `getValidationChecks`, `putValidationChecks`, `tryValidationCheck`

### Categories

The groups agents are sorted into on the Agents page. Every company starts with a set.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Add, rename or remove categories
- Preview what a change moves

In Falcon: Categories

In the API: `previewCompanyCategoryChange`, `applyCompanyCategoryChange`

### Models and spend limits

Choose which AI models agents may use, and set daily and monthly spend limits, so a runaway loop can't run up a bill.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Approve models for the company
- Set daily and monthly spend limits
- See each model's use against its limit

In Falcon: Models & limits

In the API: `actOnCompanyModelLimits`, `listModelChoices`, `getAgentDailyCostCap`

### Bring your own model keys

Use your own key for a model provider, such as OpenRouter or xAI, so that spend is on your account.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.
>
> Available since 29 September 2026.

What you can do:

- Add a provider key
- Choose which models use it

In Falcon: Models & limits

In the API: `setCompanyModelsSecret`

### More model providers

Use OpenAI directly, any OpenAI-compatible service, or your company's own Amazon Bedrock.

> Coming next: Designed and planned, not built yet.

What you can do:

- Connect OpenAI or a compatible service
- Use models from your own Bedrock account

### Billing and plan

See the company's plan, keep payment methods, and download invoices.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- See the current plan
- Add or change a payment method
- Download invoices

In Falcon: Billing & plan · Plan · Payment methods

In the API: `getCompanyBilling`, `getCompanyPlan`, `listCompanyPaymentMethod`, `createCompanyPaymentMethod`, `createBillingPortalSession`, `downloadBillingInvoice`

### Audit log

A record of who changed what, and when, across the company.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Search the log
- Export it

In Falcon: Audit log

In the API: `listCompanyAudit`, `searchCompanyAudit`, `exportCompanyAudit`

### Data and privacy

Answer people's requests to see or delete their data, and export what the company has written.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Record and complete a privacy request
- Export the company's authored content

In Falcon: Data & privacy

In the API: `actOnCompanyDataPrivacy`, `listCompanyPrivacyRequests`, `createCompanyPrivacyRequest`, `completeCompanyPrivacyRequest`, `exportCompanyAuthoredContent`

### Deployment accounts

Run your agents in Falcon's AWS account, or in your company's own.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.
>
> Runs in: your own AWS account.

What you can do:

- Add your own AWS account
- See which account each agent runs in

In Falcon: Deployment accounts

In the API: `listCompanyDeploymentAccount`, `createCompanyDeploymentAccount`, `getCompanyDeploymentDetails`

## Who can do what

Who may do each thing in this guide, and what an AI assistant connected as that person may do with it. Roles are set in Company settings › Roles.

| What | Who may | An assistant |
|---|---|---|
| Reads the company's theme: the live one and any draft | Company administrators | Reads · needs “look” |
| Changes the draft theme | Company administrators | Changes · needs “change settings” · waits for your approval |
| Makes a preview of the draft theme | Company administrators | Changes · needs “change settings” · waits for your approval |
| Applies the draft theme so everyone sees it | Company administrators | Changes · needs “change settings” · waits for your approval |
| Goes back to an earlier theme | Company administrators | Changes · needs “change settings” · waits for your approval |
| Uploads a logo for the theme | Company administrators | Changes · needs “change settings” · waits for your approval |
| Lists the company's agent images | Company administrators | Reads · needs “look” |
| Adds an agent image | Company administrators | Changes · needs “change settings” · waits for your approval |
| Lists the people in the company | Company administrators | Reads · needs “look” |
| Searches the company's people | Company administrators | Reads · needs “look” |
| Invites a person to the company | Company administrators | Reaches outside · needs “change settings” · waits for your approval |
| Changes a person's role or details | Company administrators | Changes · needs “change settings” · waits for your approval |
| Signs a person out everywhere | Company administrators | Cannot be undone · needs “change settings” · waits for your approval |
| Preview or invite many people | Company administrators | Reaches outside · needs “change settings” · waits for your approval |
| List roles | Anyone who can see it | Reads · needs “look” |
| Run a security action | Company administrators | Changes · needs “change settings” · waits for your approval |
| Run a single sign-on action | Company administrators | Changes · needs “change settings” · waits for your approval |
| Read Microsoft sign-in settings | Company administrators | Reads · needs “look” |
| Save Microsoft sign-in settings | Company administrators | Changes · needs “change settings” · waits for your approval |
| Record Microsoft admin approval | Company administrators | Changes · needs “change settings” · waits for your approval |
| Read company validation checks | Company administrators | Reads · needs “look” |
| Save company validation checks | Company administrators | Changes · needs “change settings” · waits for your approval |
| Try a validation check | Company administrators | Changes · needs “change settings” · waits for your approval |
| Shows what moving agents or instructions between categories would change | Company administrators | Reads · needs “look” |
| Moves agents or instructions between categories, using a preview you made just before | Company administrators | Changes · needs “change settings” · waits for your approval |
| Run a model settings action | Company administrators | Changes · needs “change settings” · waits for your approval |
| Lists the models you can choose for an agent | Anyone who can see it | Reads · needs “look” |
| Get agent daily cost cap | Anyone who can see it | Reads · needs “look” |
| Save a model credential | Company administrators | Changes · needs “change settings” · waits for your approval |
| Read billing details | Company administrators | Reads · needs “look” |
| Read the company plan | Company administrators | Reads · needs “look” |
| Lists the company's payment methods | Company administrators | Reads · needs “look” |
| Adds a payment method | Company administrators | Changes · needs “change settings” · waits for your approval |
| Open the billing portal | Company administrators | Reaches outside · needs “change settings” · waits for your approval |
| Download billing invoice | Company administrators | Reads · needs “look” |
| Lists entries in the company's audit log | Company administrators | Reads · needs “look” |
| Searches the company's audit log | Company administrators | Reads · needs “look” |
| Export the audit log | Company administrators | Changes · needs “change settings” · waits for your approval |
| Run a data privacy action | Company administrators | Changes · needs “change settings” · waits for your approval |
| List privacy requests | Company administrators | Reads · needs “look” |
| Record a privacy request | Company administrators | Changes · needs “change settings” · waits for your approval |
| Complete a privacy request | Company administrators | Cannot be undone · needs “change settings” · waits for your approval |
| Export authored content | Company administrators | Changes · needs “change settings” · waits for your approval |
| Lists the cloud accounts the company deploys agents to | Company administrators | Reads · needs “look” |
| Adds a cloud account to deploy agents to | Company administrators | Changes · needs “change settings” · waits for your approval |
| Read deployment details | Company administrators | Reads · needs “look” |

## Related

- Next: [Reports](https://falcon.com/docs/reports.md)
- [Glossary](https://falcon.com/docs/glossary.md)

---

Source: https://falcon.com/docs/reports

# Reports

Counts and costs for agents and workflows, never what people said.

Reports show how much agents and workflows are used and what they cost. They show counts and money, not the words of conversations.

You can export any report, or have it sent on a schedule.

## Features

### Agent reports

How much each agent is used, across the company or for one agent.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Compare agents side by side
- Open one agent's report

In Falcon: Reports · Agents · One agent

### Workflow reports

How often each workflow ran, and how often it finished, failed or waited for a person.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- See runs and outcomes for each workflow

In Falcon: Workflows

In the API: `getPipelineReport`

### Cost reports

What the company spent on models and services, by agent, workflow and day.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- See spend by agent and by workflow
- See spend against the limits you set

In Falcon: Costs

In the API: `getCostReport`

### Export or schedule a report

Download a report as a file, or have it sent on a schedule.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Export a report
- Schedule a report, change the schedule or cancel it

In Falcon: Export or schedule

In the API: `exportReport`, `getReportSchedules`, `createReportSchedule`, `updateReportSchedule`, `cancelReportSchedule`

### How closely each agent is watched

Choose, for each agent, how much detail Falcon keeps about its work.

> Coming next: Designed and planned, not built yet.

What you can do:

- Set a level for each agent

## Who can do what

Who may do each thing in this guide, and what an AI assistant connected as that person may do with it. Roles are set in Company settings › Roles.

| What | Who may | An assistant |
|---|---|---|
| Get workflow report | Anyone who can see it | Reads · needs “look” |
| Get cost report | Anyone who can see it | Reads · needs “look” |
| Export report | People who can edit it | Reaches outside · needs “change settings” · waits for your approval |
| Get report schedules | Anyone who can see it | Reads · needs “look” |
| Create report schedule | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Update report schedule | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Cancel report schedule | People who can delete it | Cannot be undone · needs “change settings” · waits for your approval |

## Related

- Next: [Ask Falcon](https://falcon.com/docs/ask-studio.md)
- [Glossary](https://falcon.com/docs/glossary.md)

---

Source: https://falcon.com/docs/ask-studio

# Ask Falcon

An assistant on every page that explains what you are looking at and can make changes for you, with your say-so.

Ask Falcon is a panel you can open on any page. Ask it what something means, or ask it to do something.

It works as you, through the same MCP tools anyone can use, so it can only do what you are allowed to do.

When it wants to change something, it shows you the change first. Nothing happens until you say yes.

## Features

### Ask Falcon

Open the panel on any page, ask a question in your own words, and get an answer about that page.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Ask what a page or setting does
- Ask it to make a change, and review the change before it is applied
- Edit or skip a proposed change

### Skills and house rules

Ask Falcon follows a short playbook for each part of the Studio, and your company can add its own house rules and reference pages for it.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- See the playbooks Ask Falcon follows and which always apply
- Add house rules for your people, for every request or one kind of job
- Give Ask Falcon reference pages about your company to look things up in

In Falcon: Ask Studio

In the API: `getAskStudioSkills`, `updateAskStudioSkills`

### Describe it

Describe an agent or a workflow in your own words, answer a few questions, and Falcon drafts it for you.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Start from a description instead of a blank page
- Answer follow-up questions
- Accept, edit or discard the draft

In Falcon: New agent · Workflows

In the API: `createGuideSession`, `answerGuideSession`, `editGuideSession`, `acceptGuideSession`, `discardGuideSession`

## Who can do what

Who may do each thing in this guide, and what an AI assistant connected as that person may do with it. Roles are set in Company settings › Roles.

| What | Who may | An assistant |
|---|---|---|
| Read Ask Studio's skills | Anyone who can see it | Reads · needs “look” |
| Change Ask Studio's house rules | People who manage skills | Changes · needs “change settings” |
| Create guide session | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Answer guide session | People who can edit it | Changes · needs “run and test” · waits for your approval |
| Edit guide session | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Accept guide session | People who can edit it | Changes · needs “change settings” · waits for your approval |
| Discard guide session | People who can edit it | Changes · needs “change settings” · waits for your approval |

## Related

- Next: [MCP and API](https://falcon.com/docs/mcp-and-api.md)
- [Glossary](https://falcon.com/docs/glossary.md)

---

Source: https://falcon.com/docs/mcp-and-api

# MCP and API

Everything you can do in Falcon, an AI assistant or your own code can do too, within your access.

What you do in Falcon is also available over MCP, the protocol AI assistants use to call tools. Every build checks that the two match.

An assistant acts as the person who connected it. It can only do what that person may do, and risky changes wait for their approval.

The same operations are available as an HTTP API. The API reference lists every operation, grouped by area, and marks the few that assistants cannot call, such as sign-in steps that need a browser.

## Features

### Connect an AI assistant

Connect an AI assistant to Falcon over MCP and sign in as yourself.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Connect an assistant with a browser sign-in
- See connected assistants and disconnect any of them

In Falcon: MCP · Connected assistants · Connect an app

In the API: `authorizeMcpClient`, `reviewMcpConsent`, `listMcpConnections`, `revokeMcpConnection`

### Approve what an assistant asks to do

When an assistant asks to do something risky, such as publish or send, it waits until you approve it in Falcon.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Review a pending request with its details
- Approve or cancel it

In Falcon: Approvals · Review an app’s request

In the API: `getMcpPendingOperation`, `confirmMcpPendingOperation`, `cancelMcpPendingOperation`

### Company MCP policy

Administrators decide whether people may connect assistants, and what those assistants may do.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Allow or stop assistant connections
- Limit what connected assistants may do

In Falcon: MCP

In the API: `getCompanyMcpPolicy`, `updateCompanyMcpPolicy`

### HTTP API

The same operations as MCP, over HTTP, each described with when to use it and who may call it.

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

What you can do:

- Read the API reference
- Call operations as a signed-in person

In the API: `getMe`, `getCompany`

### Falcon inside your own product

Show parts of Falcon inside your own app, so your users can shape their agents there.

> Coming next: Designed and planned, not built yet.

What you can do:

- Embed an agent's settings in your app

## Who can do what

Who may do each thing in this guide, and what an AI assistant connected as that person may do with it. Roles are set in Company settings › Roles.

| What | Who may | An assistant |
|---|---|---|
| Authorize an MCP connection | The browser holding the required sign-in, setup, enrollment or consent challenge; callbacks also require their bound state. | Not offered to assistants |
| Review an MCP connection request | The browser holding the required sign-in, setup, enrollment or consent challenge; callbacks also require their bound state. | Not offered to assistants |
| List your MCP connections | Anyone who can see it | Reads · needs “look” |
| Disconnect an MCP client | People who can edit it | Cannot be undone · needs “change settings” · waits for your approval |
| Read an MCP request awaiting review | The browser holding the required sign-in, setup, enrollment or consent challenge; callbacks also require their bound state. | Not offered to assistants |
| Confirm an MCP request | The browser holding the required sign-in, setup, enrollment or consent challenge; callbacks also require their bound state. | Not offered to assistants |
| Cancel an MCP request | The browser holding the required sign-in, setup, enrollment or consent challenge; callbacks also require their bound state. | Not offered to assistants |
| Read MCP access policy | Company administrators | Reads · needs “look” |
| Change MCP access policy | Company administrators | Changes · needs “change settings” · waits for your approval |
| Get me | Anyone who can see it | Reads · needs “look” |
| Get company | Anyone who can see it | Reads · needs “look” |

## Related

- Next: [Every feature](https://falcon.com/docs/features.md)
- [Glossary](https://falcon.com/docs/glossary.md)

---

Source: https://falcon.com/docs/features

# Every feature

107 features: 0 live, 90 in pilot and 17 coming next.

- **Live** (0): Built and running in Falcon today.
- **In pilot** (90): Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.
- **Coming next** (17): Designed and planned, not built yet.

## [Sign-in and your account](https://falcon.com/docs/account.md)

- **Sign in with an email code** (In pilot): Type your email address and enter the code Falcon sends you. There is no password to remember or leak.
- **Authenticator apps and passkeys** (In pilot): Add an authenticator app or a passkey as another way to sign in, with backup codes for when you lose a device.
- **Sign in with Google or Microsoft** (In pilot): An administrator can let people sign in with their Google or Microsoft work account. Email codes stay available as a way back in.
- **Set up the workspace** (In pilot): The administrator named when the workspace was created signs in once to claim it. Everyone else joins by invitation.
- **Join by invitation** (In pilot): People join a company in Falcon only when someone with the right role invites them. The invitation says which role they get.
- **Your profile, notifications and appearance** (In pilot): Keep your name and photo current, choose which notices you get, and pick light or dark.
- **Verify your phone number** (In pilot): Add a mobile number to your profile and confirm it with a code sent by text.
- **Phone codes for two-step** (In pilot): Use a code sent to your mobile by text, RCS or a call to finish signing in and to confirm it's you. Passkeys and authenticator apps stay the stronger choice.
- **Ask to join a company** (Coming next): Someone with a company email address can ask to join, and an administrator says yes or no.

## [Agents](https://falcon.com/docs/agents.md)

- **The Agents page** (In pilot): See every agent you can open, grouped by category, with whether each is a draft, Live or waiting for approval.
- **Start a new agent** (In pilot): Start from a blank agent or a starter for a common job, then change anything you like.
- **Instructions** (In pilot): Tell the agent who it is, what it is for and how to behave, in plain words, with a preview of what it will be told.
- **Skills** (In pilot): A skill is a reusable piece of know-how, with a note on when to use it. Write one once and share it across agents.
- **Knowledge** (In pilot): Give agents your help articles, policies and documents to answer from. Falcon keeps its copy up to date.
- **Tools** (In pilot): Tools let an agent do things, such as look something up, save a file or send a message. You choose which tools each agent may use.
- **Your own endpoints and MCP servers as tools** (In pilot): Turn your own web endpoint or a remote MCP server into a tool your agents can call.
- **Messaging tools** (In pilot): Tools for sending texts, WhatsApp, Slack, Telegram and email, with company rules for each.
- **Memory** (In pilot): Choose what an agent remembers about the people it talks to, and send what it learns to your own systems.
- **Variables** (In pilot): Give an agent named values, such as an opening time or a booking link, that its instructions can use.
- **Choose the model** (In pilot): Pick which AI model an agent uses from the models your company allows.
- **Voice** (In pilot): Choose whether an agent is off on calls, only listens and takes notes, or listens and talks.
- **Channels** (In pilot): Choose where people can reach an agent: a chat widget on your website, a link, texts and RCS, WhatsApp, Slack, Telegram or email.
- **Ask an admin** (In pilot): When an agent is unsure, it asks a person instead of guessing, and tells them only what they need to answer.
- **Who an agent's messages come from** (In pilot): Choose the identity an agent speaks as by default, such as a person on your team or one of your places.
- **Test an agent** (In pilot): Talk to the draft before anyone else does, look inside each reply, and save good conversations as test cases.
- **Make a version Live** (In pilot): Nothing reaches customers until someone with permission makes a version Live. Falcon checks the agent is ready first.
- **Versions** (In pilot): Every published version is kept, so you can see what changed between two of them.
- **Access and collaborators** (In pilot): Decide who can see, edit or publish each agent, and invite collaborators to one agent only.
- **Built on this** (In pilot): Let other agents build on this one, so a change here reaches them too, and see what a change would affect.
- **Activity** (In pilot): See what an agent has been doing. The words of a conversation stay hidden until someone with permission chooses to show them.
- **Pause, copy, hand over or archive** (In pilot): Pause an agent, copy it, hand it to someone else, archive it or take its content with you.
- **Session rules** (Coming next): Rules for how long a conversation lasts and when a new one begins.
- **Script tools** (Coming next): Write a short script as a tool, with its own settings and secrets, and let agents call it.
- **Rate replies and improve them** (Coming next): Mark a reply good or bad and ask Falcon to suggest a change to the instructions.
- **Push notifications** (Coming next): Reach people through notifications on their phone, as another channel.

## [Identities](https://falcon.com/docs/identities.md)

- **Company identities** (In pilot): Keep the company's people and places in one list, each with its own page and a short code you can use to refer to it.
- **Your own identities** (In pilot): Keep identities of your own, for agents you build for yourself.
- **Identity fields** (In pilot): Decide once which details every identity has. Name, company, communication style, background and voice are built in; you can hide them or add your own.
- **Checks on identity details** (In pilot): Falcon checks what is typed into an identity's fields, such as whether a name looks like a name, before it is saved.
- **Working hours** (In pilot): Give a person or place its hours, so workflows know when it is open.
- **Where an identity can be reached** (In pilot): Each identity an agent acts as can have its own email address, phone number, voice address, link and website chat widget.
- **RCS for each identity** (In pilot): Messages can come from the company's shared RCS sender, or an identity can have its own.
- **An identity's own mailbox and calendar** (In pilot): Connect an identity's own mailbox and calendar, so an agent can send as them and check when they are free.
- **Images of an identity** (In pilot): Add photos of a person or place, with a note on when an agent should use each one.
- **Import identities** (In pilot): Bring in many identities at once from a file. Falcon shows every row before anything is saved.
- **How the company's email looks** (In pilot): Set the layout and logo placement for email your agents send, once for the whole company.
- **Outlook mailboxes and calendars** (Coming next): Connect an identity's Outlook mailbox and Microsoft calendar, as with Gmail.
- **A customer's contact points** (Coming next): See every way a customer has reached you, which agent and identity answered, and when.
- **Pictures drawn in an identity's likeness** (Coming next): Let an agent draw a picture of an identity from its reference images, labelled as made by AI.

## [Workflows](https://falcon.com/docs/workflows.md)

- **Workflows and how they work** (In pilot): See the company's workflows, and read a short guide to how a workflow runs.
- **Edit a workflow** (In pilot): Add, move and remove steps on a canvas, with the chosen step's settings beside it.
- **How a workflow starts** (In pilot): Start a workflow on a schedule, when an app sends an event, when a conversation arrives, or when another system calls it.
- **Side chains and retries** (In pilot): Hang extra steps off a step for when someone is interested, replies STOP, or the step fails.
- **Hours check** (In pilot): A step that checks whether a person or place is open now, and sends the workflow one way or the other.
- **Front desk** (In pilot): A first step that greets the person, looks them up in past conversations, and hands them to the right agent, such as sales or service.
- **Engagements** (In pilot): Say what counts as someone engaging, such as replying or booking, so each can start a side chain.
- **Waiting for you** (In pilot): Drafts that need a person before they are sent, from every workflow, in one place.
- **Request mapping** (In pilot): Give any outside system its own address to call, and map what it sends into a workflow or an agent.
- **Runs** (In pilot): See each time a workflow ran, and step in when one needs it.
- **Publish and roll back** (In pilot): Make a workflow Live, go back to an earlier version, or archive it.
- **Webhooks out** (In pilot): Tell your own systems when a workflow does something, by sending them a signed message.
- **When a reply could be for more than one** (In pilot): If two agents or identities wrote to someone and they reply, choose whether the reply goes to the latest, to both, or is read to decide.
- **Taking turns** (In pilot): Share new conversations between identities in turn, such as sales reps who are available, and keep returning people with the same one.
- **Acting as on a step** (In pilot): Pick which identity an agent acts as on each step, so the same agent can appear three times as three people.
- **Topics and keywords on messages** (Coming next): Label outgoing messages with a topic, so opting out of marketing never stops order updates, and route replies by keyword.
- **Evaluate steps** (Coming next): A step that scores or checks the work of earlier steps, and workflows that only evaluate.
- **Sequences that repeat** (Coming next): Repeat a set of steps, such as a follow-up every few days, until someone replies or a limit is reached.
- **Nudges and review rounds** (Coming next): Remind people who haven't answered, and pass work through more than one review before it goes out.

## [Connections](https://falcon.com/docs/connections.md)

- **App connections** (In pilot): Connect business apps such as a CRM, payments or email with a browser sign-in, using Falcon's sign-in or your own.
- **Your connections** (In pilot): Connections that belong to you, such as your own mailbox, and what may use them.
- **Email sending** (In pilot): Send your agents' email from your own domain, through SendGrid, or through any mailbox that offers SMTP and IMAP.
- **Twilio** (In pilot): Connect your Twilio account to send and receive texts with your own numbers.
- **Phone numbers and AI voicemail** (In pilot): Give an identity its own phone number for incoming calls, and let calls outside hours go to an AI voicemail that takes a message.
- **Storage and files** (In pilot): Connect an Amazon S3 bucket so agents can save, read and list files, and share files by link.
- **Event sources** (In pilot): Let a system send events to Falcon, so a workflow can start when something happens there.
- **Domains and links** (In pilot): Set which domains the company's links and addresses use.
- **GitHub repositories** (In pilot): Install the GitHub app for your organisation and choose which repositories agents may work in.
- **Your own REST API** (Coming next): Describe your own REST API once and turn its endpoints into tools.
- **Message keys** (Coming next): Keys that let your own systems send messages through your agents' channels.

## [Company settings](https://falcon.com/docs/company-settings.md)

- **General and brand** (In pilot): The company's name, logo, colours and images, previewed before you apply them.
- **People** (In pilot): See everyone in the company, invite someone new, and make a person inactive when they leave.
- **Invite many people at once** (In pilot): Upload a list of people with their roles and send all the invitations together.
- **Roles** (In pilot): A role is a set of things a person may do. Use the built-in roles or make your own.
- **Security** (In pilot): Decide how people sign in, whether a second step is required, and how long sessions last.
- **Single sign-on** (In pilot): Let people sign in through the company's own Microsoft tenant or identity provider.
- **Checks** (In pilot): One place for the checks Falcon runs on what people type, such as identity details, with a way to try each one.
- **Categories** (In pilot): The groups agents are sorted into on the Agents page. Every company starts with a set.
- **Models and spend limits** (In pilot): Choose which AI models agents may use, and set daily and monthly spend limits, so a runaway loop can't run up a bill.
- **Bring your own model keys** (In pilot): Use your own key for a model provider, such as OpenRouter or xAI, so that spend is on your account.
- **More model providers** (Coming next): Use OpenAI directly, any OpenAI-compatible service, or your company's own Amazon Bedrock.
- **Billing and plan** (In pilot): See the company's plan, keep payment methods, and download invoices.
- **Audit log** (In pilot): A record of who changed what, and when, across the company.
- **Data and privacy** (In pilot): Answer people's requests to see or delete their data, and export what the company has written.
- **Deployment accounts** (In pilot): Run your agents in Falcon's AWS account, or in your company's own.

## [Reports](https://falcon.com/docs/reports.md)

- **Agent reports** (In pilot): How much each agent is used, across the company or for one agent.
- **Workflow reports** (In pilot): How often each workflow ran, and how often it finished, failed or waited for a person.
- **Cost reports** (In pilot): What the company spent on models and services, by agent, workflow and day.
- **Export or schedule a report** (In pilot): Download a report as a file, or have it sent on a schedule.
- **How closely each agent is watched** (Coming next): Choose, for each agent, how much detail Falcon keeps about its work.

## [Ask Falcon](https://falcon.com/docs/ask-studio.md)

- **Ask Falcon** (In pilot): Open the panel on any page, ask a question in your own words, and get an answer about that page.
- **Skills and house rules** (In pilot): Ask Falcon follows a short playbook for each part of the Studio, and your company can add its own house rules and reference pages for it.
- **Describe it** (In pilot): Describe an agent or a workflow in your own words, answer a few questions, and Falcon drafts it for you.

## [MCP and API](https://falcon.com/docs/mcp-and-api.md)

- **Connect an AI assistant** (In pilot): Connect an AI assistant to Falcon over MCP and sign in as yourself.
- **Approve what an assistant asks to do** (In pilot): When an assistant asks to do something risky, such as publish or send, it waits until you approve it in Falcon.
- **Company MCP policy** (In pilot): Administrators decide whether people may connect assistants, and what those assistants may do.
- **HTTP API** (In pilot): The same operations as MCP, over HTTP, each described with when to use it and who may call it.
- **Falcon inside your own product** (Coming next): Show parts of Falcon inside your own app, so your users can shape their agents there.

---

Source: https://falcon.com/docs/glossary

# Glossary

The words Falcon uses, in plain language, in alphabetical order.

Each term links to the guide that explains it in full. The API and MCP keep a few older names: a workflow is a `pipeline` there.

**Agent**: A named assistant that talks to your customers or does work for your team. You shape it in plain words: instructions, skills, knowledge, tools, memory, voice and channels. See [Agents](https://falcon.com/docs/agents).

**Built on**: An agent can build on another agent, so a change to the one underneath reaches every agent built on it. Falcon shows which agents a change would affect before you make it. See [Built on this](https://falcon.com/docs/agents#built-on).

**Check**: A test Falcon runs on what people type, such as an identity's details, before it is saved or used. A workflow step can also be a check instead of an agent. See [Checks](https://falcon.com/docs/company-settings#checks).

**Draft**: The version of an agent or workflow that people are still changing. Customers never meet a draft; you can test it at any time. See [Make a version Live](https://falcon.com/docs/agents#publishing).

**Engagement**: What your company counts as someone engaging, such as replying or booking a meeting. Each engagement can start a side chain. See [Engagements](https://falcon.com/docs/workflows#engagements).

**Identity**: A person or a place an agent acts for, such as Dana in sales or the Downtown store. Identities are records your company keeps; they never sign in and are not users. See [Identities](https://falcon.com/docs/identities).

**Live**: The version of an agent or workflow that customers meet and that runs for real. Only someone with permission makes a version Live, and you can go back to an earlier one. See [Publish and roll back](https://falcon.com/docs/workflows#workflow-publish).

**Side chain**: Extra steps that hang off one step of a workflow and run only when something happens: someone is interested, replies STOP, or the step fails. See [Side chains and retries](https://falcon.com/docs/workflows#side-chains).

**Start**: What makes a workflow run: a schedule, an event from a connected app, a conversation arriving, another system calling it, or another workflow failing. A workflow can have more than one start. See [How a workflow starts](https://falcon.com/docs/workflows#workflow-starts).

**Waiting for you**: One place for the drafts that need a person before they are sent, from every workflow. You send a draft as it is, edit it first, or decide not to send it. See [Waiting for you](https://falcon.com/docs/workflows#waiting-for-you).

**Workflow**: A series of steps that runs when something happens, such as a new email, a form, a call or a time of day. Each step is an agent or a check. Like agents, workflows have drafts and a Live version. See [Workflows](https://falcon.com/docs/workflows).

## Related

- [Getting started](https://falcon.com/docs)
- [Every feature](https://falcon.com/docs/features)
- [Troubleshooting](https://falcon.com/docs/troubleshooting)

---

Source: https://falcon.com/docs/choose-how-to-run

# Choose how to run Falcon

Run your agents with Hosted by Falcon, or in your company's own AWS account. You build them the same way in both.

Every company starts with Hosted by Falcon. You build, test and make agents Live the same way in both, so the choice is about where they run and who AWS bills, not about how you work.

## Compare the two

|  | Hosted by Falcon | Your own AWS account |
|---|---|---|
| What you set up | Nothing. Sign up and build. | An AWS account for your company: one Falcon creates for you, or one you already own. You add a role it trusts. |
| Where agents run | On Amazon Bedrock AgentCore, in Falcon's AWS account, kept apart from every other company. | On Amazon Bedrock AgentCore, in your AWS account. Each environment can have its own account. |
| Who AWS bills | Falcon, which bills you for what your agents use. | You, directly, for what runs in your account. Falcon bills you for Falcon. |
| Your data | Kept in Falcon's AWS account, in storage for your company alone. | Kept in your account, where your own AWS tools and policies apply. |
| Spend protection | Daily and monthly limits, and a switch that stops model calls at once. | The same limits, plus anything you set in AWS, such as budgets. |
| How Falcon reaches it | Falcon runs it for you. | Through roles your account trusts, a few minutes at a time: read-only checks, and deploy access only when someone deploys. Deleting the role cuts access at once. |
| If Falcon can't reach it | Falcon looks after it. | Your agents keep running. Falcon holds no AWS keys. |
| Best for | Getting started, pilots and most companies. | Companies whose rules say data and compute stay in their own cloud account. |

## What stays the same

- Building, testing and making versions Live, in the same pages.
- Who can do what: roles and permissions come from Falcon, not from AWS.
- Connections, identities and workflows.
- Using your own model keys, so model spend is on your provider account.
- AI assistants connected over MCP, with the same access as the person who connected them.

## Before you choose your own account

> In pilot: Built, but not yet on for everyone: it is being tried out with pilot companies, waiting on a provider, or still to be released.

An administrator adds the account in **Company settings › Deployment accounts**. Falcon shows which account each agent runs in, and when it last checked and deployed each one.

## Related

- [Deployment accounts](https://falcon.com/docs/company-settings#deployment-accounts)
- [Models and spend limits](https://falcon.com/docs/company-settings#models-and-limits)
- [Bring your own model keys](https://falcon.com/docs/company-settings#own-model-keys)

---

Source: https://falcon.com/docs/connect-your-assistant

# Connect your assistant

Use Falcon from Claude, ChatGPT, Cursor, VS Code or another AI assistant over MCP. It signs in as you and can do only what you can do.

## Before you start

- An administrator turns assistants on in the **MCP** section, and chooses who may connect one.
- Copy your **server address** from **MCP › Connect**. It is your company's address followed by `/mcp`, such as `https://studio.example.com/mcp`.
- Your assistant must be one your company has set up. An administrator can tell you which, and give you a client ID if your assistant asks for one.
- There is no key to paste. The assistant opens your browser, you sign in, and you choose what it may do.

## Claude Code

Run this in a terminal, then type `/mcp` in Claude Code and choose the server to sign in.

```sh
claude mcp add --transport http my-studio https://studio.example.com/mcp
```

## Claude Desktop and claude.ai

- Open **Settings › Connectors** and choose **Add custom connector**.
- Paste your server address. If an administrator gave you a client ID, enter it under **Advanced settings**.
- Sign in when Claude asks.

## ChatGPT

- Open **Settings › Apps & Connectors**. Under **Advanced settings**, turn on **Developer mode**.
- Choose **Create**, give the connector a name, paste your server address and choose **OAuth**.
- Sign in when ChatGPT asks.

## Cursor

Add this to `~/.cursor/mcp.json`, or to `.cursor/mcp.json` in one project, then sign in from **Cursor Settings › MCP**.

```json
{
  "mcpServers": {
    "my-studio": {
      "url": "https://studio.example.com/mcp"
    }
  }
}
```

## VS Code

Add this to `.vscode/mcp.json` in your project, or run **MCP: Add Server** from the command palette and choose HTTP. Start the server from the file and sign in when asked.

```json
{
  "servers": {
    "my-studio": {
      "type": "http",
      "url": "https://studio.example.com/mcp"
    }
  }
}
```

## Codex and other assistants

For Codex, run these two commands. The second signs you in.

```sh
codex mcp add my-studio --url https://studio.example.com/mcp
codex mcp login my-studio
```

Most other assistants take the same settings as Cursor. **MCP › Connect** shows them with your address filled in.

## What your assistant sees

Many assistants load only a few dozen tools, and every tool costs them attention. So an assistant is offered a small set that still reaches everything:

| Tool | What it does |
|---|---|
| `ask_studio` | Ask in plain words. It reads what it needs as you, then answers or suggests changes, each showing what it would change. Nothing changes until you apply a suggestion. |
| `find_operations` | Find what your assistant can do by what it does. |
| `describe_operation` | One operation's description, the access it needs and exactly what it takes. |
| `run_operation` | Run any operation by name, with the same checks as doing it yourself. |
| `builder_list_agents` | List the agents you can see. |
| `builder_get_agent_run` | Check how a run went. |
| `builder_search_docs` | Search these docs. |
| `builder_read_doc_page` | Read one docs page in full. |

For automation that wants exact calls, an administrator can switch on the tools for whole areas, such as Agents or Workflows, for everyone or for one assistant.

## What it may do

- When you sign in, you choose what the assistant may do: **look**, **change settings**, and **run and test**.
- It can never do more than you can. It gets what you allowed and your role allows, whichever is less.
- Risky changes, such as making a version Live, deleting or anything that sends, wait until you approve them in Falcon.
- Everything it does is in the audit log as you, through that assistant, for example “Dana via Claude Desktop”.
- Disconnect it at any time from **Connections › Yours**. Its access ends at once.

## Your Live agents and workflows as tools

> Coming next: Designed and planned, not built yet.

You will be able to offer a Live agent or workflow to connected assistants as a tool of its own, switched on per agent or workflow, with the same spend limits and approvals as any other channel.

## Related

- [MCP and API guide](https://falcon.com/docs/mcp-and-api)
- [Approve what an assistant asks to do](https://falcon.com/docs/mcp-and-api#assistant-approvals)
- [API and MCP reference](https://falcon.com/docs/api)
- [Troubleshooting](https://falcon.com/docs/troubleshooting)

---

Source: https://falcon.com/docs/examples

# Worked examples

Real requests and payloads for webhooks out, request maps and the failure handler, ready to copy.

Every example on this page is what the API takes and sends, with made-up ids. Addresses such as `company.example.invalid` stand for your own.

## Webhooks out

Tell your own system when a workflow finishes or fails. Add a webhook to the workflow with the address to call and the events it wants. In the API a workflow is a `pipeline`:

```http
POST /api/v0/pipelines/pipeline_webhook_fixture/webhooks
Idempotency-Key: example-request-343

{
  "name": "Notify our system",
  "destination": "webhook",
  "address": "https://company.example.invalid/pipeline-events",
  "events": [
    "falcon.pipeline.run.finished"
  ],
  "include_results": false,
  "enabled": false
}
```

When a run finishes, the address receives one signed request. The body is a CloudEvent; `data` says how the run went, step by step:

```http
POST https://company.example.invalid/pipeline-events
content-type: application/cloudevents+json
webhook-id: evt_2b7d9e41
webhook-timestamp: 1790690539
webhook-signature: v1,tADkis+ZUs0deVI6jKf4G56u24Ohuz1R4mQQe3Yv6ag=

{
  "specversion": "1.0",
  "type": "falcon.pipeline.run.finished",
  "source": "/pipelines/pipeline_new_leads",
  "id": "evt_2b7d9e41",
  "subject": "run_4f9c2a",
  "time": "2026-09-29T14:02:19.000Z",
  "datacontenttype": "application/json",
  "data": {
    "run_id": "run_4f9c2a",
    "request_id": "req_7Hc2xQ",
    "pipeline": "pipeline_new_leads",
    "pipeline_version": "version_12",
    "outcome": "finished",
    "started_at": "2026-09-29T14:02:11.000Z",
    "finished_at": "2026-09-29T14:02:19.000Z",
    "steps": [
      {
        "step": "qualify",
        "outcome": "finished"
      },
      {
        "step": "reply",
        "outcome": "finished"
      }
    ]
  }
}
```

### Check the signature

Webhooks are signed the Standard Webhooks way. Join `webhook-id`, `webhook-timestamp` and the exact body with dots, sign that with HMAC-SHA256 using the webhook's secret (the part after `whsec_`, base64-decoded), and compare it with the value after `v1,` in `webhook-signature`. Refuse a timestamp more than five minutes old.

```js
import { createHmac, timingSafeEqual } from "node:crypto";

function verify(secret, headers, body) {
  const key = Buffer.from(secret.replace(/^whsec_/, ""), "base64");
  const signed = `${headers["webhook-id"]}.${headers["webhook-timestamp"]}.${body}`;
  const expected = createHmac("sha256", key).update(signed).digest();
  return headers["webhook-signature"]
    .split(" ")
    .map((part) => Buffer.from(part.replace(/^v1,/, ""), "base64"))
    .some((given) => given.length === expected.length && timingSafeEqual(given, expected));
}
```

A failed run sends `falcon.pipeline.run.failed`, and its `data` adds the step that failed and why. Answer with any 2xx status within ten seconds. Anything else is tried again after 30 seconds, 2 minutes, 10 minutes, then 15 minutes twice, and then marked as not delivered in the webhook's delivery log.

## Request maps

Give another system its own address to call, and say which parts of what it sends go into the workflow. This map checks a shared key and takes the order number from the body:

```http
POST /api/v0/connections/incoming-requests
Idempotency-Key: incoming-map-example

{
  "name": "order_paid",
  "target": {
    "kind": "workflow",
    "workflow_id": "workflow_fixture"
  },
  "acting_as": {
    "kind": "agent_rule"
  },
  "topic": null,
  "keywords": [],
  "person_fields": [],
  "verification": {
    "kind": "shared_key",
    "header": "X-Falcon-Key",
    "key": "fixture-key"
  },
  "conditions": [],
  "mapping": [
    {
      "path": "order.id",
      "input": "order_id",
      "transform": "text"
    }
  ],
  "idempotency_field": "id"
}
```

The answer gives the map its address. The key itself is never shown again: `secret_set` only says there is one.

```json
{
  "id": "request_map_fixture",
  "name": "order_paid",
  "target": {
    "kind": "workflow",
    "workflow_id": "workflow_fixture"
  },
  "acting_as": {
    "kind": "agent_rule"
  },
  "topic": null,
  "keywords": [],
  "person_fields": [],
  "verification": {
    "kind": "shared_key",
    "header": "X-Falcon-Key",
    "secret_set": true
  },
  "conditions": [],
  "mapping": [
    {
      "path": "order.id",
      "input": "order_id",
      "transform": "text"
    }
  ],
  "idempotency_field": "id",
  "address": "/v1/request-maps/example",
  "last_request": null,
  "recent_results": [],
  "revision": "1",
  "needs_look": false,
  "target_version": null
}
```

Your system then calls the address with the key in the header the map names. When the workflow is still running, the answer says it was accepted:

```http
POST /v1/request-maps/example
X-Falcon-Key: ••••••••

{
  "id": "event_fixture"
}
```

```http
HTTP/1.1 202

{
  "request_id": "request_fixture",
  "state": "accepted",
  "run_id": "run_fixture",
  "received_at": "2026-09-28T00:00:00.000Z"
}
```

When the workflow answers quickly, the answer carries its result and where to read the run:

```http
HTTP/1.1 200

{
  "request_id": "request_fixture",
  "state": "finished",
  "run_id": "run_fixture",
  "status_url": "/api/v0/pipelines/pipeline_fixture/runs/run_fixture",
  "result": {
    "text": "Your order is on its way."
  },
  "received_at": "2026-09-28T00:00:00.000Z"
}
```

## The failure handler

Choose one workflow to run whenever another workflow fails or stops for someone to look at it. Give it the start **A workflow failed or needs attention**, and narrow it by workflow, category or outcome if you like. Then make it the company's handler in **Company settings › General › When a workflow fails**, or with the API:

```http
PUT /api/v0/settings/workflow-failures
If-Match: "1"

{
  "handler_pipeline_id": "pipeline_tell_the_team"
}
```

A workflow can name its own handler instead, in its settings. The handler starts with exactly what a webhook out sends for the failed run:

```json
{
  "run_id": "run_8d31e0",
  "request_id": "req_Qm41Lp",
  "pipeline": "pipeline_new_leads",
  "pipeline_version": "version_12",
  "outcome": "not_done",
  "started_at": "2026-09-29T15:40:02.000Z",
  "finished_at": "2026-09-29T15:40:31.000Z",
  "steps": [
    {
      "step": "qualify",
      "outcome": "finished"
    },
    {
      "step": "reply",
      "outcome": "failed"
    }
  ],
  "failed_step": "reply",
  "error": {
    "code": "dispatch_failed"
  }
}
```

- `outcome` is `not_done`. `failed_step` is left out when the run stopped as a whole rather than at one step.
- A handler's own runs never start a handler, and test runs never do.
- The handler starts once for each failed run.

## Related

- [Webhooks out](https://falcon.com/docs/workflows#webhooks-out)
- [Request mapping](https://falcon.com/docs/workflows#request-mapping)
- [How a workflow starts](https://falcon.com/docs/workflows#workflow-starts)
- [Troubleshooting](https://falcon.com/docs/troubleshooting)

---

Source: https://falcon.com/docs/troubleshooting

# Troubleshooting

What each problem code means and what to do about it, with the exact title the error shows.

When something is refused, the answer is a problem with a `code`, a `title` and a `request_id`. Find the code on this page. Every answer also carries an `X-Request-Id` header with the same id: give it to support and they can find exactly what happened.

```json
{
  "type": "about:blank",
  "title": "Permission denied",
  "status": 403,
  "code": "permission_denied",
  "request_id": "req_7Hc2xQ"
}
```

| Code | Status | What it means | What to do |
|---|---|---|---|
| `invalid_request` | 400 Invalid request | Something in the request is missing or not in the right form. | Check the fields against the API reference. The problem's `field` and `reason`, when present, say which one. |
| `unauthenticated` | 401 Unauthenticated | You are not signed in, or your sign-in has ended. | Sign in again. An assistant connected over MCP signs in again through your browser. |
| `permission_denied` | 403 Permission denied | You are signed in, but your role does not allow this, or an assistant's access does not include it. | Ask an administrator to check your role in Company settings › Roles. For an assistant, reconnect it and allow the access it needs. |
| `step_up_required` | 403 step up required | This change needs you to confirm it's you first, with your second step. | Confirm with your authenticator app, passkey or code, then try again. |
| `not_found` | 404 Not found | What you asked for does not exist in this company, or you cannot see it. | List what is there now and use one of those. Someone may have deleted or moved it. |
| `unknown_host` | 404 Unknown host | The address you used does not belong to a company. | Check your company's address, the one you sign in at. |
| `idempotency_conflict` | 409 Idempotency conflict | The same request key was used before for a different request. | Use a new `Idempotency-Key` for a new request. Reuse a key only to repeat exactly the same request. |
| `in_use` | 409 in use | This cannot be deleted because something still uses it. | The problem's `used_by` lists what uses it that you can open. Change or remove those first. |
| `reply_route_unavailable` | 409 reply route unavailable | The way to send this reply is not available right now. | Check the connection it uses under Connections, then try again. |
| `stale_revision` | 412 Stale revision | Someone changed this after you read it. | Read it again, look at what changed, then make your change on the new version. |
| `reply_route_denied` | 422 reply route denied | A reply was sent to an address this agent may not reply from. | Check which addresses the agent may use in its channels. |
| `sender_field_refused` | 422 sender field refused | A sender or identity field was refused, for example because it could be mistaken for someone else. | The problem's `field` and `reason` say which field and why. Change that field and save again. |
| `precondition_required` | 428 Precondition required | A change needs to say which version it changes. | Read the current version and send its revision in the `If-Match` header. |
| `rate_limited` | 429 rate limited | Too many requests in a short time. | Wait a little and try again. The `Retry-After` header says how long, when it is sent. |
| `usage_limit_reached` | 429 The monthly limit for Ask Studio and MCP calls is reached, or an administrator has paused them. An administrator can change this under Reports › Studio assistant and MCP. | A monthly cap for you or for this assistant has been reached, or an administrator paused assistant use. | Ask an administrator to raise the cap or turn assistant use back on in Reports › Costs. |
| `internal_error` | 500 Internal error | Something went wrong on our side. Nothing was changed unless the result says so. | Check the current state before you try a change again, and contact support with the request id. |
| `not_implemented` | 501 not implemented | This is designed but not built on this host yet. | Check the feature's status on its guide. Coming next features answer this until they are built. |
| `provider_unavailable` | 503 provider unavailable | A service this depends on is not available right now, or it is not switched on for your company. | Try again later. If it keeps happening, contact support with the request id. |
| `auth_failed` |  | A mailbox or service refused the sign-in details for a connection. | Check the connection's details under Connections and connect it again. |
| `email_unverified` |  | The email address has not been confirmed yet. | Open the confirmation email and follow its link, then try again. |
| `identity_code_taken` |  | Another identity already uses this code. | Choose a different code. |
| `identity_code_unbound` |  | This identity code is not linked to anyone yet. | Link it on the identity's page first. |
| `identity_code_unknown` |  | No identity has this code. | Check the code on the identity's page. |
| `imap_auth_failed` |  | The mailbox refused the sign-in for reading mail. | Check the mailbox user name and password, or app password, and connect it again. |
| `imap_unavailable` |  | The mailbox could not be reached for reading mail. | Check the mailbox server address and port, then try again later. |
| `likeness_refused` |  | A picture or video tool refused: the model tried to choose a picture, look, model or seed; a real person's likeness has no current consent; or a rating does not fit. | The problem's `field` and `reason` say which. Record consent on the identity's Likeness, pick a look that fits the rating, or leave pictures to the look and identity. |
| `relay_denied` |  | The mail server would not send this message on your behalf. | Check the sending settings with whoever runs the mail server. |
| `tls_required` |  | A mail server was reached without a secure connection. | Use the server's secure port or turn on TLS in the connection. |

## Related

- [API and MCP reference](https://falcon.com/docs/api)
- [Connect your assistant](https://falcon.com/docs/connect-your-assistant)
- [Glossary](https://falcon.com/docs/glossary)

---

Source: https://falcon.com/docs/api

# API and MCP reference

562 operations in 27 areas. 513 of them are available to AI assistants over MCP.

Every operation runs as a signed-in person, with that person’s access. An assistant connected over MCP is limited in the same way, and risky changes wait for the person’s approval in Falcon. Operations marked API only are steps a browser has to take itself, such as signing in, or addresses other services call.

## Agent authoring and personal settings

Create and configure agents, review their work, manage invitations and maintain your personal settings.

- **Accept guide session** `acceptGuideSession`
- **Accepttransfer agent** `accepttransferAgent`
- **Agenda add** `agenda_add`
- **Agenda list** `agenda_list`
- **Agenda update** `agenda_update`
- **Answer guide session** `answerGuideSession`
- **Approves or rejects a tool that is waiting for review on an agent** `approveAgentTool`
- **Approves or rejects the changes waiting for review in one category of an agent** `reviewAgentCategory`
- **Archive agent** `archiveAgent`
- **Asks for access to an agent you can see but not open** `requestAgentInvitation`
- **Cancel background task** `cancelBackgroundTask`
- **Change an agent's chat window skin** `putAgentWidgetSkin`: Replace the whole chat window skin of one agent at once: its colours in light and dark, font, corners, launcher, header and whether the powered-by line shows.
- **Changes an agent's instructions** `updateAgentInstructions`
- **Changes an agent's settings, voice and model** `updateAgentSettings`
- **Changes an agent's variables** `updateAgentVariables`
- **Changes how an agent writes on each channel** `updateAgentChannelStyles`
- **Changes the company's settings for the agenda tool** `putAgendaSettings`
- **Changes what an agent remembers and how** `updateAgentMemory`
- **Changes which agent this agent is built on** `updateAgentBuiltOn`
- **Changes which tools and knowledge searches an agent has** `updateAgentTools`
- **Changes who can see and edit an agent** `updateAgentAccess`
- **Changes your own preferences** `updateUserPreferences`
- **Changes your own profile** `updateUserProfile`
- **Compares two versions of an agent and shows what changed** `compareAgentVersions`
- **Create agent from template** `createAgentFromTemplate`
- **Create agent receiving address** `createAgentReceivingAddress`
- **Create agent test run** `createAgentTestRun`
- **Create guide session** `createGuideSession`
- **Create memory webhook** `createMemoryWebhook`
- **Create public link invitation** `createPublicLinkInvitation`
- **Creates a new agent** `createAgent`
- **Decline agent transfer** `declineAgentTransfer`
- **Delete an agent** `deleteAgent`: Delete an agent for good.
- **Discard guide session** `discardGuideSession`
- **Disconnect connected app** `disconnectConnectedApp`
- **Download billing invoice** `downloadBillingInvoice`
- **Duplicate agent** `duplicateAgent`
- **Edit guide session** `editGuideSession`
- **Export agent content** `exportAgentContent`
- **Exports staff, community members or the audit log as CSV** `exportCompanyCsv`
- **Get company** `getCompany`: Read the public company identity for the current host, including before sign-in.
- **Get guide session** `getGuideSession`
- **Get guide target** `getGuideTarget`
- **Get me** `getMe`
- **Import public link invitations** `importPublicLinkInvitations`
- **Inspect agent test run** `inspectAgentTestRun`
- **Invites a person to work on an agent** `inviteAgentCollaborator`
- **List connected apps** `listConnectedApps`
- **List memory deliveries** `listMemoryDeliveries`
- **List people** `listPeople`
- **List policies** `listPolicies`
- **List public link invitations** `listPublicLinkInvitations`
- **List roles** `listRoles`
- **Lists the agents a new agent can be built on** `listAgentBases`
- **Lists the agents you can see, with the ones you can open and the ones you can only see the name of** `listAgents`
- **Lists the models you can choose for an agent** `listModelChoices`
- **Pause agent** `pauseAgent`
- **Publishes the agent's current draft as a new version** `publishAgent`
- **Put agent channels** `putAgentChannels`
- **Put agent public entry** `putAgentPublicEntry`
- **Put agent voice settings** `putAgentVoiceSettings`
- **Reads your own preferences** `getUserPreferences`
- **Reads your own profile** `getUserProfile`
- **Reads your own sign-in methods and sessions** `getUserSecurity`
- **Removes your own access to an agent someone shared with you** `leaveSharedAgent`
- **Replay conversation steps** `replayConversationSteps`
- **Resend public link invitation** `resendPublicLinkInvitation`
- **Restore agent** `restoreAgent`
- **Reveal conversation words** `revealConversationWords`: reveal Conversation Words Requires a reason and records an audit entry before revealing private conversation words.
- **Revoke public link invitation** `revokePublicLinkInvitation`
- **Save agent test case** `saveAgentTestCase`
- **Service disconnect connected app** `serviceDisconnectConnectedApp`
- **Service start connected app** `serviceStartConnectedApp`
- **Service test connected app event** `serviceTestConnectedAppEvent`
- **Shows a preview of the agent on one channel (chat, texts, email, WhatsApp or voice)** `getStudioPreview`
- **Signs you out of one session, or everywhere** `revokeUserSession`
- **Test agent phone channel** `testAgentPhoneChannel`
- **Transfer agent** `transferAgent`
- **Update agent knowledge searches** `updateAgentKnowledgeSearches`
- **Update agent tool settings** `updateAgentToolSettings`

## Agents

Inspect agent configuration, inheritance, publication readiness and channel settings before making changes.

- **List a person's memory records** `listAgentMemoryRecords`: List what an agent keeps in memory about one person: each record's kind, whose memory it is ('The person, everywhere' or 'The person, with each identity'), the path it is kept under and when it was kept.
- **List agent dependencies** `listAgentDependencies`: Read the dependencies reported for this agent.
- **List agent images** `listAgentImages`: Read the image records available on the agent and their revision.
- **List agent test cases** `listAgentTestCases`: Read the agent's saved test cases and current revision.
- **List agent test runs** `listAgentTestRuns`: Read recorded test-run summaries and their states.
- **List saved agent versions** `listAgentVersions`: List version numbers, authors and the environments using each version.
- **Read a person's effective agent access** `getEffectiveAgentAccess`: Read resolved permissions, category and section scope for the named person on this agent.
- **Read agent email and phone channels** `getAgentChannels`: Read phone and email channel configuration, enabled state and readiness reasons.
- **Read agent instruction inheritance** `getAgentInheritance`: Read the base chain, inherited parts and insertion points visible to the caller.
- **Read agent instruction sections** `listAgentSections`: Read grouped instruction sections and their editable state.
- **Read agent instructions** `getAgentInstructions`: Read the instruction content and revision visible to the caller.
- **Read agent knowledge searches** `getAgentKnowledgeSearches`: Read configured knowledge searches, source filters, mode and enabled state.
- **Read agent lifecycle options** `getAgentLifecycle`: Read lifecycle state and the actions currently described for this agent.
- **Read agent memory settings** `getAgentMemory`: Read memory kinds, retention, processing steps, sharing destinations and lookup-tool settings, who shares each kind of memory ('The person, everywhere', 'The person, with each identity', 'A case' or 'The whole agent') and the path each kind is kept under.
- **Read agent public entry settings** `getAgentPublicEntry`: Read public link, widget and app entry settings, usage limits and reported readiness.
- **Read agent publication readiness** `getAgentPublish`: Read the current publication state and readiness information before publishing.
- **Read agent settings** `getAgentSettings`: Read model selection, token and budget settings, input options and current revision.
- **Read agent sharing settings** `getAgentAccess`: Read the agent access configuration and current revision under the caller's existing authority.
- **Read agent test choices** `getAgentTestChoices`: Read the choices offered for configuring an agent test.
- **Read agent variable definitions** `getAgentVariables`: Read the agent variable configuration and its revision.
- **Read agent voice settings** `getAgentVoiceSettings`: Read voice configuration and reported limits, usage and readiness for the selected environment.
- **Read agent writing styles by channel** `getAgentChannelStyles`: Read channel-specific instruction styles and their placement relative to shared sections.
- **Read an agent** `getAgent`: Read the agent name, description, category, visibility and current revision.
- **Read an agent base selection** `getAgentBuiltOn`: Read the selected base agent and inheritance settings.
- **Read an agent overview** `getAgentOverview`: Read the agent overview with the state and summary information available to the caller.
- **Read an agent tool settings layer** `getAgentToolSettings`: Read the settings and argument defaults explicitly configured for this agent and tool.
- **Read an agent's chat window skin** `getAgentWidgetSkin`: Read the live chat window skin of one agent: its colours in light and dark, font, corners, launcher, header, whether the powered-by line shows, and whether the plan lets you hide it.
- **Read effective agent tool settings** `getAgentEffectiveToolSettings`: Read merged tool settings with each value's source and lock state, plus the resulting model input schema.
- **Read skills bound to an agent** `getAgentSkills`: Read bound skill references, versions and per-agent usage guidance and tool switches.
- **Read the impact of an agent base** `getBuiltOnImpact`: Read the impact information for the agent's base relationship.
- **Read tools assigned to an agent** `getAgentTools`: Read the tools assigned to this agent, their guidance and approval state.

## AI guidance

Inspect published AI guidance.

- **List AI guidance** `listFalconGuidance`: List AI guidance with the full text, declared variables and usage locations.
- **List saved guidance samples** `listFalconGuidanceSamples`: Read example sample inputs and expected labels for this published guide.
- **Propose an AI guidance draft** `proposeFalconGuidanceDraft`: Replace the registered content parts of a draft.
- **Publish an AI guidance draft** `publishFalconGuidanceDraft` (API only): Publish the exact independently reviewed draft revision from the staff console with a fresh second step.
- **Read a guidance trial** `getFalconGuidanceTrial`: Read a saved trial with the exact source guide and sample revision, answers and confidence.
- **Read AI guidance and its history** `getFalconGuidance`: Read AI guidance and its history with the full text, declared variables and usage locations.
- **Read an AI guidance draft** `getFalconGuidanceDraft`: Read the current draft and its review state without changing published guidance.
- **Review an AI guidance draft** `reviewFalconGuidanceDraft`: Approve the exact draft revision as a staff member who contributed none of its edits.
- **Save a guidance sample** `saveFalconGuidanceSample`: Save a example sample and expected labels with its current revision.
- **Try AI guidance on a saved sample** `runFalconGuidanceTrial`: Run the current publication, an earlier published version or an exact saved draft revision against one saved example sample.

## Approvals where people work

Choose where each kind of approval is asked — Studio, Slack, email, and later Microsoft Teams and text — and who is asked, and see who answered and where.

- **Change where approvals are asked** `updateApprovalRouting`: Choose, for each approval rule, where it is asked (Studio is always included) and name the approvers asked outside Studio.
- **List approvals asked outside Studio** `listApprovalAsks`: List the approvals asked in Slack or by email that you were asked or asked for: where they went, whether they were sent or recorded in test mode, who answered and how.
- **Read where approvals are asked** `getApprovalRouting`: Read, for each approval rule, where it is asked and who is asked, and which channels are ready, switched off, need setting up or are designed only.

## Ask Studio skills

See the playbooks Ask Studio follows for each part of the Studio, and add your company's own house rules and reference pages for it.

- **Change Ask Studio's house rules** `updateAskStudioSkills`: Replace the company's Ask Studio house rules and reference pages.
- **Read Ask Studio's skills** `getAskStudioSkills`: Read Ask Studio's built-in skills (name, when each applies, its intent and whether it always applies), with the company's house rules and reference pages.

## Browser authentication and consent

Sign in, verify security challenges and review connected MCP clients.

- **Accept a membership invitation** `acceptInvitation` (API only): Accept the invitation for the verified recipient.
- **Authorize an MCP connection** `authorizeMcpClient` (API only): Grant a registered MCP client the permissions the person approved in the browser.
- **Cancel an MCP request** `cancelMcpPendingOperation` (API only): Cancel a pending request without executing it.
- **Choose how phone codes are sent** `setPhoneTwoStepDelivery` (API only): Choose text, RCS (where offered) or a voice call for the current person's phone codes.
- **Confirm an MCP request** `confirmMcpPendingOperation` (API only): Approve execution of the exact pending request in its original browser review session.
- **Confirm the code and turn on phone two-step** `confirmPhoneTwoStep` (API only): Confirm the code sent to the new number.
- **Disconnect an MCP client** `revokeMcpConnection`: Revoke the selected MCP connection and its continued access.
- **Finish connecting a preset** `completeConnectorPreset` (API only): Contract and synthetic fixture only.
- **Finish connecting an app** `serviceCompleteConnectedApp` (API only): Company/person-bound single-use callback state.
- **Finish signing in** `completeSignIn` (API only): Exchange the challenge response for a browser session.
- **Finish workspace setup** `completeWorkspaceSetup` (API only): Save the workspace’s initial configuration after the setup challenge.
- **Hand a provider answer back to its company** `relaySignIn` (API only): At the shared sign-in address, finish a Google or Microsoft answer for a sign-in started at another company address and return where to hand it back.
- **List your MCP connections** `listMcpConnections`: Read the MCP clients connected by the current person.
- **Read a membership invitation** `getInvitation` (API only): Look up a membership invitation using its invitation proof.
- **Read an MCP request awaiting review** `getMcpPendingOperation` (API only): Read the exact pending request in the original browser review session.
- **Read an uncertain MCP request’s receipt** `reconcileMcpPendingOperation` (API only): Read an execution receipt without repeating the uncertain action.
- **Read sign-in choices** `getSignInOptions` (API only): Read the sign-in methods available for this workspace.
- **Read workspace setup** `getWorkspaceSetup` (API only): Read the remaining setup required for this workspace.
- **Remove a security method** `removeSecurityMethod`: Remove a security method after the required browser security check.
- **Replace backup codes** `regenerateBackupCodes`: Invalidate the old backup codes and issue replacements for the current person.
- **Reset a person’s security methods** `resetPersonFactors`: Reset the selected person’s security methods under current administrator authority.
- **Review an MCP connection request** `reviewMcpConsent` (API only): Validate the registered client, redirect and requested permissions before browser consent.
- **Send a code to add a phone for two-step** `startPhoneTwoStep` (API only): Send a six-digit code to a mobile number the current person is adding as a second step.
- **Send a phone code for a second step** `sendSecondStepPhoneCode` (API only): Send a code to the current person's verified number for this sign-in or Confirm it's you flow.
- **Send a phone verification code** `startPhoneVerification` (API only): Send a verification challenge to the current person’s phone.
- **Start a security check** `startStepUp` (API only): Creates a ten-minute browser/session-bound challenge.
- **Start adding a security method** `beginSecurityEnrollment` (API only): Create a challenge to enroll a passkey or another supported security method.
- **Start signing in** `startSignIn` (API only): Create a sign-in challenge for the selected method.
- **Verify a new security method** `confirmSecurityEnrollment` (API only): Verify the enrollment response and attach the security method to the current person.
- **Verify a phone number** `confirmPhoneVerification` (API only): Verify the phone challenge and save the verified phone.
- **Verify a second sign-in step** `completeSecondStep` (API only): Complete the current session’s second authentication challenge.

## Branding

Work on the company's white-label look with your own assistant: the Studio theme (colours, type, logos, the sign-in page and email) and each agent's chat window skin. Every change is a draft you can preview, check for contrast and compare with live before a person publishes it in Studio.

- **Change an agent's draft skin** `updateBrandingSkinDraft`: Change the draft chat window skin of one agent, never the live one.
- **Change the draft theme** `updateBrandingThemeDraft`: Change the draft theme, never the live one.
- **Check the branding's contrast** `getBrandingContrast`: Check the draft (or live) theme and an agent's skin against WCAG 2.1 AA: every text colour against every surface it sits on, in light and dark, with its measured ratio.
- **Compare the draft branding with live** `compareBranding`: List what the draft changes against live: the tokens and logos you set, every drawn value that follows from them in light and dark, and, with an agent, each chat window setting that changes.
- **Discard an agent's draft skin** `discardBrandingSkinDraft`: Throw an agent's draft skin away.
- **Discard the draft theme** `discardBrandingThemeDraft`: Throw the draft theme away.
- **Get a signed link to upload a logo** `createBrandingLogoUpload`: Get a short-lived, single-use link to upload a logo for light or dark.
- **Preview the draft branding** `previewBranding`: Preview the draft theme and an agent's draft skin without changing anything live: returns a signed Studio link that shows the draft beside live in light and dark, what changes, the contrast result, and screenshots of the sign-in page, the agents list, an agent's page, the chat window and an email in light and dark as images.
- **Publish an agent's draft skin** `publishBrandingSkin`: Make an agent's draft skin live on every site that embeds its chat window.
- **Publish the draft theme** `publishBrandingTheme`: Make the draft theme live for everyone in the company: Studio, the sign-in page and company email.
- **Read a branding preview's pages** `getBrandingPreview`: Read the pages of a preview made by previewBranding: each screen in light and dark, drafted and live, as self-contained pages with no scripts, plus what changes.
- **Read an agent's chat window skin** `getBrandingSkin`: Read one agent's chat window skin: what is live, the draft being worked on (if any) and earlier live versions.
- **Read the company theme as design tokens** `getBrandingTheme`: Read the company's Studio theme: the live tokens and logos, the draft being worked on (if any), both as DTCG design tokens for light and dark, every token a company may set with its default, and earlier live versions to roll back to.
- **Roll an agent's live skin back** `rollbackBrandingSkin`: Make an earlier live skin, from history, live again.
- **Roll the live theme back** `rollbackBrandingTheme`: Make an earlier live theme, from history, live again.
- **Upload a logo to a signed link** `uploadBrandingLogo` (API only): Send a logo's bytes to the link createBrandingLogoUpload gave.

## Change review

See exactly what changed in an agent, skill or workflow before it is published, and set the company's rules for changes to chosen categories and for reviewing workflows.

- **Approve or decline a workflow review** `decidePipelineReview`: Approve or decline a workflow's review.
- **Approve or reject a skill draft for publishing** `reviewSkillPublication`: Record a second person's decision on a skill draft whose category needs approval before publishing.
- **Ask for a review of a draft** `requestChangeReview`: Tell the people who watch an agent or skill, and those who can approve it, that its draft is ready for review, with a short note.
- **Change the company's change rules** `updateChangeRules`: Choose which instruction categories need an acknowledgement or a second person's approval before publishing, which always alert their owners, and when a workflow needs a review before it is published.
- **Comment on a workflow review** `commentPipelineReview`: Add a comment to a workflow review, on one step or start or on the whole workflow.
- **List your change alerts** `listChangeAlerts`: List alerts about agents, skills and categories you watch, newest first, each with what changed and the diff as you may see it.
- **Mark a change alert read** `markChangeAlertRead`: Mark one of your change alerts as read.
- **Read a workflow's review** `getPipelineReview`: Read whether a workflow needs a review before it is published, the latest review with its comments, who may review it, and the diff it is judged by.
- **Read the company's change rules** `getChangeRules`: Read which instruction categories need an acknowledgement or a second person's approval before publishing, and which always alert their owners.
- **Read whether you are told about changes** `getChangeWatch`: Read whether you are told about changes to an agent, a skill or an instruction category, and why: your own choice, a default for category owners and administrators, or a company rule.
- **Send a workflow for review** `requestPipelineReview`: Ask a named person to review a workflow's draft before it is published.
- **Show what changed in a skill** `getSkillDiff`: Compare a skill's draft or a published version with Live, with nothing, or with another published version, line by line.
- **Show what changed in a workflow** `getPipelineDiff`: Compare a workflow's draft with its latest version, Live, nothing or a published version: each step and start line by line, and an 'Also changes' list of the agent versions it follows, connections, variables and limits.
- **Show what changed in an agent** `getAgentDiff`: Compare an agent's draft or a published version with Live, with nothing, or with another published version, line by line.
- **Turn change alerts on or off** `updateChangeWatch`: Choose whether you are told about changes to an agent, a skill or an instruction category.

## Company administration

Manage company membership, security, appearance, billing and operating settings.

- **Accept company setup suggestions** `companyAcceptGuideSession`: Apply the reviewed company setup guide result under the current caller’s settings authority and declared preconditions.
- **Add a company person** `createCompanyPerson`: Create a membership record with the supplied name, address and role.
- **Add a GitHub installation** `createCompanyGithubInstallation`: Record a GitHub installation and its repository access for this company.
- **Adds a cloud account to deploy agents to** `createCompanyDeploymentAccount`
- **Adds a payment method** `createCompanyPaymentMethod`
- **Adds an agent image** `createCompanyImage`
- **Answer a company setup question** `companyAnswerGuideSession`: Submit an answer to the current company setup guide question and return the next guide state.
- **Applies the draft theme so everyone sees it** `changeCompanyThemeApply`
- **Change billing details** `updateCompanyBilling`: Save the company’s billing details using their current revision.
- **Change email domain policy** `updateCompanyDomainPolicy`: Replace the allowed and blocked email domain policy using its current revision.
- **Change MCP access policy** `updateCompanyMcpPolicy`: Replace the company MCP policy with reviewed values using its current revision.
- **Changes a deployment account** `updateCompanyDeploymentAccount`
- **Changes a payment method** `updateCompanyPaymentMethod`
- **Changes a person's role or details** `updateCompanyPerson`
- **Changes an agent image** `updateCompanyImage`
- **Changes an invitation** `updateCompanyInvitation`
- **Changes the defaults for traveller payments** `updateTravellerPaymentDefaults`
- **Changes the draft theme** `changeCompanyThemeDraft`
- **Complete a privacy request** `completeCompanyPrivacyRequest`: Mark or execute the requested completion under the privacy request’s current state and company policy.
- **Copies a pending invitation's link** `copyCompanyInvitationLink`: Returns the link of one pending invitation so an administrator can pass it on directly, for example when its email could not be sent.
- **Discard a company setup guide** `companyDiscardGuideSession`: Discard the proposed company setup guide result without applying its suggested settings.
- **Edit a company setup answer** `companyEditGuideSession`: Change a saved answer in the company setup guide before accepting its proposed settings.
- **Export a company package** `exportCompanyPackage`: Make one package of this company's checks, identity fields, categories, company skills, agents and workflows, with the list of connections, numbers and identities the company importing it must supply.
- **Export authored content** `exportCompanyAuthoredContent`: Request an export of authored content visible under the company’s current access policy.
- **Export the audit log** `exportCompanyAudit`: Request an export of company audit records under the caller’s current audit permissions.
- **Goes back to an earlier theme** `changeCompanyThemeRevert`
- **Import a company package** `importCompanyPackage`: Import a package as its preview said: add new items, update the items an earlier import of the same package made, and leave out refused items.
- **Invites a person to the company** `createCompanyInvitation`
- **List privacy requests** `listCompanyPrivacyRequests`: List the company’s recorded privacy requests and their progress so authorized staff can handle them.
- **Lists earlier themes** `listCompanyThemeHistory`
- **Lists entries in the company's audit log** `listCompanyAudit`
- **Lists the choices available for agent images** `getCompanyImageOptions`
- **Lists the cloud accounts the company deploys agents to** `listCompanyDeploymentAccount`
- **Lists the company's agent images** `listCompanyImage`
- **Lists the company's invitations** `listCompanyInvitation`
- **Lists the company's payment methods** `listCompanyPaymentMethod`
- **Lists the design tokens a theme can change** `getCompanyThemeTokens`
- **Lists the filters and choices for the people list** `getCompanyPeopleOptions`
- **Lists the people in the company** `listCompanyPerson`
- **Makes a preview of the draft theme** `changeCompanyThemePreview`
- **Moderate company authors** `moderateCompanyAuthors`: Apply the requested moderation action to the selected community authors.
- **Moves agents or instructions between categories, using a preview you made just before** `applyCompanyCategoryChange`
- **Open the billing portal** `createBillingPortalSession`: Create a short-lived billing portal session for the company so an authorized person can manage payment details with the billing provider.
- **Preview a company package import** `previewCompanyPackageImport`: Check every item in a package against this company: which it would add, which it would update because an earlier import (or an item with the same name or key) made them, which are unchanged, and which it would leave out and why.
- **Preview or invite many people** `invitePeople`: Preview up to 500 invitation rows, then queue ready invitations after one recent second step.
- **Read a company setup guide** `companyGetGuideSession`: Read the current question, answers and proposed settings for an existing company setup guide session.
- **Read billing details** `getCompanyBilling`: Read company billing configuration and available invoice information.
- **Read company message topics** `getCompanyMessageTopics`: Company message topics and their consent classification, quiet hours and daily limits.
- **Read company validation checks** `getValidationChecks`: Read company-defined identity and input validation checks.
- **Read deployment details** `getCompanyDeploymentDetails`: Read the deployment details for the supplied account and resource selection.
- **Read email domain policy** `getCompanyDomainPolicy`: Read the company rules for allowed and blocked email domains before changing enrollment or invitation policy.
- **Read MCP access policy** `getCompanyMcpPolicy`: Read who may connect MCP clients and which actions require confirmation under the current company policy.
- **Read Microsoft sign-in settings** `getCompanyMicrosoftSignIn`
- **Read the company plan** `getCompanyPlan`: Read the current company plan and its limits before requesting a feature that depends on plan eligibility.
- **Reads one agent image** `getCompanyImage`
- **Reads one deployment account** `getCompanyDeploymentAccount`
- **Reads one invitation** `getCompanyInvitation`
- **Reads one payment method** `getCompanyPaymentMethod`
- **Reads one person** `getCompanyPerson`
- **Reads payment totals and defaults for traveller payments** `getTravellerPaymentTotals`
- **Reads the company's theme: the live one and any draft** `getCompanyTheme`
- **Reads the full details for one person** `getCompanyPersonDetails`
- **Record a privacy request** `createCompanyPrivacyRequest`: Record a person’s requested privacy action for company handling.
- **Record Microsoft admin approval** `completeCompanyMicrosoftConsent`
- **Run a company action** `actOnCompanyGeneral`: Run a supported company-level action against the current settings revision.
- **Run a connection action** `actOnCompanyConnection`: Run a supported action on one company connection.
- **Run a data privacy action** `actOnCompanyDataPrivacy`: Run a supported action on the company data privacy settings.
- **Run a deployment account action** `actOnCompanyDeploymentAccount`: Run a supported action on a company deployment account.
- **Run a GitHub installation action** `actOnCompanyGithubInstallation`: Run a supported action on a company GitHub installation.
- **Run a model settings action** `actOnCompanyModelLimits`: Run a supported action on company model settings using the current revision.
- **Run a security action** `actOnCompanySecurity`: Run a supported action on company security settings.
- **Run a single sign-on action** `actOnCompanySso`: Run a supported action on the company’s single sign-on configuration.
- **Run a tool category action** `actOnCompanyToolCategory`: Run a supported action on a company tool category.
- **Run an event source action** `actOnCompanyEventSource`: Run a supported action on one company event source.
- **Run an MCP policy action** `actOnCompanyMcpPolicy`: Run a supported action on the company MCP policy, subject to current company authority and the declared revision precondition.
- **Runs an action on a payment method, such as verify or remove** `actOnCompanyPaymentMethod`
- **Runs an action on a person, such as sign out everywhere or remove. The actions allowed depend on the person** `actOnCompanyPerson`: Runs an action on a person, such as sign_out_everywhere or remove.
- **Runs an action on an agent image, such as build or remove. The actions allowed depend on the image** `actOnCompanyImage`: Runs an action on an agent image, such as build or remove.
- **Runs an action on an invitation, such as resend or revoke** `actOnCompanyInvitation`
- **Save a connection credential** `setCompanyConnectionsSecret`: Replace the write-only credential for one company connection.
- **Save a model credential** `setCompanyModelsSecret`: Replace the write-only credential in company model settings.
- **Save an event source credential** `setCompanyEventSourcesSecret`: Replace the write-only credential for one company event source.
- **Save company validation checks** `putValidationChecks`: Replace company-defined checks used with built-in safety and field-fit checks.
- **Save Microsoft sign-in settings** `putCompanyMicrosoftSignIn`
- **Searches the company's audit log** `searchCompanyAudit`
- **Searches the company's people** `searchCompanyPeople`
- **Show what a company package carries** `getCompanyPackages`: Read how many checks, identity fields, categories, company skills, agents and workflows a package from this company would carry, what never travels, and the packages imported here before.
- **Shows what moving agents or instructions between categories would change** `previewCompanyCategoryChange`
- **Signs a person out everywhere** `signOutCompanyPerson`
- **Start a company setup guide** `companyCreateGuideSession`: Create a company setup guide session for the requested settings target.
- **Throws away the draft theme** `changeCompanyThemeDiscard`
- **Try a validation check** `tryValidationCheck`: Evaluate sample text with one company-defined check and return the decision without saving the sample.
- **Update company message topics** `updateCompanyMessageTopics`: Company message topics and their consent classification, quiet hours and daily limits.
- **Uploads a logo for the theme** `uploadCompanyLogo`

## Company closing

Close a company's account from the operator console: two people agree, and 30 days later its data is removed.

- **Cancel closing a company** `cancelOperatorCompanyClosure`: Withdraw a proposal, or cancel a scheduled closing before its 30 days run out.
- **Confirm closing a company** `confirmOperatorCompanyClosure`: Confirm another staff member's proposal.
- **List company closings** `listOperatorCompanyClosures`: List every company closing that has been proposed, scheduled, withdrawn or cancelled, with who did each step and when the data goes.
- **Propose closing a company** `proposeOperatorCompanyClosure`: Propose closing a company, with a reason.

## Company self-serve sign-up

Create and review a new company workspace without granting production access before operator staff approval.

- **Approve a company sign-up** `approveCompanySignupRequest`: Approve a verified request and create its limited workspace defaults.
- **List company email reputation guards** `listEmailReputation`: Read rolling SES bounce and complaint rates and any automatic outbound-email pauses.
- **List company sign-up requests** `listCompanySignupRequests`: Read verified company sign-ups waiting for operator staff approval.
- **Start a company sign-up** `startCompanySignup` (API only): Submit company details, a verified mailbox and a processor-vault payment reference for a new workspace sign-up.
- **Unpause a company's outbound email** `unpauseCompanyEmail`: After review, staff can clear a company's automatic email-reputation pause.
- **Verify a company sign-up** `verifyCompanySignup` (API only): Verify the mailbox code sent for a company sign-up.

## Connections

Inspect configured app and channel connections, sender readiness and tool grants.

- **Change company app sign-in settings** `updateCompanyAppSettings`: Change between Falcon's client and the company's own client.
- **Change identity verification settings** `updateIdentityVerificationSettings`: Change how checks work for every agent: the age the inquiry template checks, whether a finished check counts as verified, how long links work, when people verify again, and the notice sent with every link.
- **Connect a Persona account** `saveIdentityVerificationAccount`: Connect the company's own Persona account for Dev (a sandbox key) or Live (a production key), with the inquiry template and the webhook secret.
- **Connect an Other email mailbox** `createOtherEmailConnection`: Connect a hand-entered SMTP and IMAP mailbox after a mandatory test send; credentials are write-only and encrypted.
- **Create an incoming email map** `createIncomingEmailMap`: Create an ordered email rule that maps a recipient, sender, subject or keyword to a workflow or agent.
- **Create an incoming request map** `createIncomingRequestMap`: Create a signed, bounded request address that maps an incoming request to a workflow or agent.
- **Decide on a held email** `decideHeldIncomingEmail`: Start a held or skipped email anyway (its email map's target runs once), mark it as spam, or dismiss it.
- **Delete an incoming request map** `deleteIncomingRequestMap`: Delete a request map.
- **Disconnect a Persona account** `removeIdentityVerificationAccount`: Disconnect an environment's Persona account and schedule its key and webhook secret for deletion.
- **Disconnect an Other email mailbox** `disconnectOtherEmailConnection`: Disable a mailbox and remove stored credentials.
- **List a connection's tool choices** `listToolConnectionChoices`: List tools available through a selected connection.
- **List available message senders** `listSendingSenders`: List senders and their declared channel readiness.
- **List company message keys** `listCompanyMessageKeys`: Read message-key metadata and configured state.
- **List connected app deliveries** `serviceListConnectedAppDeliveries`: Inspect recorded deliveries for an app connection.
- **List connected apps** `serviceListConnectedApps`: List app connections visible in the current company scope.
- **List connection presets** `listConnectorPresets`: List reusable connection presets.
- **List held incoming email** `listHeldIncomingEmails`: Read the emails the check before a run held or skipped: each email's kind, how sure the check was, its email map, its files kept and not kept, and when it arrived and expires.
- **List incoming email maps** `listIncomingEmailMaps`: Read the ordered rules that route inbound email, the addresses assigned to agents and identities, and the company's forwarding or MX setup.
- **List incoming request maps** `listIncomingRequestMaps`: Read request maps, their rotatable addresses and recent accepted, ignored or refused results.
- **List labelled incoming email** `listIncomingEmailLabels`: Read what people said held and skipped emails really were, next to what the check said, for checking the filter later.
- **List SendGrid connections** `listSendGridConnection`: List configured SendGrid connections and their status.
- **List twilio connections** `listTwilioConnections`: List configured Twilio connections and their status.
- **List what request map targets receive and give back** `listIncomingRequestTargets`: Read every workflow or agent a request map can start: its declared start fields, the steps that receive them first, what it gives back, and JSON Schemas for its input and response generated from its definition and versioned with it.
- **Read a connected app** `serviceGetConnectedApp`: Read an existing app connection's label, state and configuration.
- **Read a connection preset** `getConnectorPreset`: Read a preset's provider and connection requirements.
- **Read a person's verification** `getPersonVerification`: Read what one person's identity check came to in Dev or Live: its status, the age it confirmed, when it was verified and expires, and Persona's inquiry id.
- **Read a SendGrid connection** `getSendGridConnection`: Inspect a SendGrid connection's configured state.
- **Read a Twilio connection** `getTwilioConnection`: Inspect a Twilio connection's configured state.
- **Read an app event example** `getAppCloudEventExample`: Inspect the app event envelope.
- **Read company app sign-in settings** `getCompanyAppSettings`: Read the sign-in choice and identity-connectability flag.
- **Read company messaging rules** `getCompanyMessagingRules`: Read the message kinds and sending rules that apply.
- **Read identity verification** `getIdentityVerification`: Read the company's Persona connection for Dev and Live (whether each is connected, its inquiry template, the key's last four characters and the webhook address to add in Persona) and how checks work: the age the template checks, whether a finished check counts, how long links work, when people verify again and the notice sent with each link.
- **Read message-key tool grants** `getMessageKeyToolGrants`: Read which tools a message key may use.
- **Read one saved Other email mailbox** `getOtherEmailConnection`: Read mailbox status and last test without credentials.
- **Read saved Other email mailboxes** `listOtherEmailConnections`: Read mailbox status, addresses and the last provider test without credentials.
- **Read the company webhook endpoint** `getCompanyWebhookEndpoint`: Inspect the configured shared webhook endpoint.
- **Receive a mapped incoming request** `receiveIncomingRequest` (API only): Verify and map a signed incoming request, sent as JSON or form fields.
- **Receive an identity verification result** `serviceReceiveIdentityVerificationEvent` (API only): Receive a Persona webhook event for the company's Dev or Live account.
- **Revoke a person's verification** `revokePersonVerification`: Take back a person's verification in Dev or Live, with a reason, and optionally ask Persona to delete the check's personal data.
- **Rotate an incoming request address** `rotateIncomingRequestMap`: Rotate a request map address while retaining its mapping and target.
- **Show a held email's excerpt** `revealHeldIncomingEmail`: Show the subject and a short excerpt of one held or skipped email.
- **Update an incoming email map** `updateIncomingEmailMap`: Change an email map's matching rule or target while retaining its inbound address.
- **Update an incoming request map** `updateIncomingRequestMap`: Change the target, verification, conditions or mapping without changing the request address.

## Docs

Search the product docs and read any page in full, so an assistant can answer how-to questions from the same pages people read.

- **Read a docs page** `readDocPage`: Read one docs page in full, as Markdown.
- **Search the docs** `searchDocs`: Find the docs sections that best match some words, best first, each with the page to read in full.

## Editing by email

Change an agent's instructions or a company skill by emailing its edit address, or email the company's Studio address for anything you could change in Studio. The Studio assistant reads the email as the sender, drafts the change and replies with it before anything is saved.

- **Change editing by email** `updateEmailEditSettings`: Turn personal edit addresses on or off, and choose whether the company address may propose any change the sender could make in Studio ('Email Studio can propose any change', on by default).
- **Confirm an emailed change** `confirmEmailEdit`: Confirm the change drafted from an email with the one-time token from its confirmation link.
- **Confirm an emailed plan** `confirmEmailEditReview` (API only): Apply the whole plan an email to the company's Studio address produced, as the person who sent it.
- **Issue an edit address** `issueEmailEditAddress`: Issue the edit address for one agent, one skill, or, naming neither, the company-wide address.
- **Read an edit address** `getEmailEditAddress`: Read the edit address for one agent, one skill, or, naming neither, the company-wide address; null until one is issued.
- **Read an emailed plan to review** `getEmailEditReview` (API only): Read the plan an email to the company's Studio address produced beyond drafts of instructions and skills: each change with its diff and, for a request with many steps, the steps Studio is drafting.
- **Read editing by email** `getEmailEditSettings`: Read whether editing by email is on, whether personal edit addresses are required, whether the company address may propose any change, the company-wide address and how many emails were dropped or refused.

## Identities

Manage people and places that agents act for.

- **Choose an identity binding's RCS sender** `updateCompanyIdentityBindingRcs`: Attach an identity RCS registration or return the binding to the agent's shared sender.
- **Choose company binding calendars** `updateCompanyIdentityBindingCalendar`: Choose which already granted calendars narrow this identity binding’s working hours.
- **Choose personal binding calendars** `updatePersonalIdentityBindingCalendar`: Choose which already granted calendars narrow this identity binding’s working hours.
- **Commit a company identity import** `commitCompanyIdentityImport`: Write reviewed rows as drafts bound to no agent, applying explicit update or skip decisions to code-then-email matches.
- **Find who a person is with** `findIdentityTurn`: Find which identity a person is with in a pool, by one of their proved contact points.
- **Move a person to another identity** `reassignIdentityTurn`: Move one person, found by a proved contact point, to another identity bound to this agent.
- **Preview a company identity import** `previewCompanyIdentityImport`: Read CSV or JSON identity rows, apply adjustable header mapping, validate every row and show code-then-email matches without writing identities or sending mail.
- **Preview an identity RCS registration** `previewCompanyIdentityBindingRcsRegistration`: Preview the RCS registration builder pre-filled from identity fields.
- **Read a company binding address test** `getCompanyIdentityBindingTest`: Test only the current receiving email, SIP address or attached number of an enabled binding.
- **Read a personal binding address test** `getPersonalIdentityBindingTest`: Test only the current receiving email, SIP address or attached number of an enabled binding.
- **Read an identity binding's RCS sender** `getCompanyIdentityBindingRcs`: Read the agent-shared or identity-dedicated RCS choice, approval status and identity sign-off.
- **Read company binding calendar status** `getCompanyIdentityBindingCalendar`: Read the calendars selected for this binding and the last successful refresh time.
- **Read personal binding calendar status** `getPersonalIdentityBindingCalendar`: Read the calendars selected for this binding and the last successful refresh time.
- **Report on taking turns** `getIdentityTurns`: New conversations per identity over the last seven days, why each identity got them (its turn, a returning person, the fallback, a switch or a move by hand), open conversations, today's count against its daily cap, and people moved between identities.
- **Request a company binding address test** `createCompanyIdentityBindingTest`: Test only the current receiving email, SIP address or attached number of an enabled binding.
- **Request a personal binding address test** `createPersonalIdentityBindingTest`: Test only the current receiving email, SIP address or attached number of an enabled binding.
- **Simulate a new conversation** `simulateIdentityTurn` (API only): Fixture mode only.

## Knowledge

Inspect knowledge bases, their sources and synchronization progress.

- **Ask about documents** `askAboutDocuments`: Ask one registered yes/no or pick-one question about referenced items without loading their content into the agent context.
- **List knowledge bases** `listKnowledgeBase`: List knowledge bases and their source summaries.
- **List knowledge sources** `listKnowledgeSource`: List sources configured for a knowledge base.
- **List knowledge syncs** `listKnowledgeSyncs`: Read recorded synchronization attempts for a knowledge base.
- **Read a knowledge base** `getKnowledgeBase`: Inspect a knowledge base's configuration, source counts and revision.
- **Read a knowledge source** `getKnowledgeSource`: Inspect a source's location, schedule, tags and current state.

## Model controls

Pause model calls and set the run ceilings every new run starts with, from the operator console.

- **Force or release the decisions fallback** `setOperatorDecisionFallback`: Skip jev everywhere and use each job's fallback, or go back to jev.
- **Pause or resume model calls** `setOperatorModelCallSwitch`: Pause or resume model calls for everyone, one provider, one company or one agent.
- **Read decision provider health** `getOperatorDecisionHealth`: Read each decisions provider's health (jev, OpenAI Decisions, the model judge): whether its circuit breaker is closed, open or probing, recent failures and latency, and whether the fallback is forced.
- **Read model-call switches and run ceilings** `getOperatorModelControls`: Read every model-call pause switch that has been set, the providers that can be paused, and the run ceilings new runs start with.
- **Set run ceilings** `setOperatorRunCeilings`: Set the turns, tool calls, tokens per turn, time and delegation depth each new run may use, up to this deployment's maximum.

## Partners

Manage consented partner access to client companies.

- **Accept a partner invitation** `acceptPartnerInvitation`: Accept a pending partner invitation as a client administrator.
- **Add partner staff** `invitePartnerStaff`: Add an existing active member of the partner company as partner staff.
- **Approve a client’s usage markup, display and service lines** `updateClientUsageTerms`
- **Approve a partner account** `approvePartnerAccount`
- **Assign staff to a client** `assignPartnerStaff`: Assign active staff to a consented client.
- **Change a partner staff role** `updatePartnerStaffRole`
- **Create a partner account** `createPartnerAccount`: Create a staff-approved partner account.
- **Export monthly client usage and margin** `exportPartnerClientSummaries`
- **Invite a partner to a client** `createPartnerInvitation`: Create a consent invitation from a partner to a client company.
- **List client assignments** `listPartnerAssignments`
- **List companies you can open** `listMySwitchableCompanies`: List the partner console and clients you are currently assigned to, at their registered Studio addresses.
- **List partner clients** `listPartnerClients`: List client companies linked to a partner.
- **List partner people in this company** `listCompanyPartnerStaff`: Show assigned partner people with the partner marker, effective role permissions and client-owned exclusion revision.
- **List partner staff** `listPartnerStaff`
- **List partners with client access** `listCompanyPartnerAccess`: Show client administrators which partners can act in this company.
- **Read a client’s agreed usage and support terms** `getClientUsageTerms`
- **Read monthly client usage and margin** `listPartnerClientSummaries`
- **Read this company’s usage at its partner’s prices** `getCompanyUsageReport`
- **Remove a client assignment** `removePartnerAssignment`
- **Remove a partner person from this company** `removeCompanyPartnerStaff`: Record a durable client-owned exclusion for one partner person.
- **Remove partner staff** `removePartnerStaff`
- **Revoke partner access** `revokePartnerAccess`: Revoke a partner's delegated access in the current client company.

## Pricing

Set the markup the operator adds to measured cost, and see cost, markup and margin per company.

- **Confirm a markup change** `confirmOperatorMarkupChange`: Confirm a markup change another member of staff proposed.
- **Export cost, markup and margin** `exportOperatorPricingMargins`: Export each company's cost, markup, price and margin for a month as CSV.
- **Propose a markup change** `proposeOperatorMarkupChange`: Propose a markup percentage for the default, a company or a partner, from a chosen date forward.
- **Read cost, markup and margin** `getOperatorPricingMargins`: Read each company's measured cost, markup, price and margin for a month, priced with the markup in force when each unit was measured.
- **Read pricing** `getOperatorPricing`: Read the default markup, each company's and partner's markup in force, every dated change and the lines that are off.
- **Set how a company sees its price** `setOperatorCompanyPriceDisplay`: Choose whether a company's billing shows cost and markup separately (itemised) or only its price (combined).
- **Withdraw a markup change** `withdrawOperatorMarkupChange`: Withdraw a markup change that is still waiting for confirmation.

## Tools, knowledge and workflow authoring

Configure reusable skills, tools, connections, knowledge, storage, workflows and reports.

- **Adds a connection to another MCP server** `createRemoteMcpConnection`
- **Adds a source to a knowledge base** `createKnowledgeSource`
- **Adds a storage connection** `createStorageConnection`
- **Adds a webhook to a workflow** `createPipelineWebhook`
- **Adds something that starts a workflow** `createPipelineTrigger`
- **Archive a workflow** `archivePipeline`: Archive a workflow you no longer use.
- **Cancel report schedule** `cancelReportSchedule`
- **Cancel workflow run** `cancelPipelineRun`
- **Changes a company skill's draft** `updateCompanySkill`
- **Changes a company tool** `updateCompanyTool`
- **Changes a connection to another MCP server** `updateRemoteMcpConnection`
- **Changes a file store** `updateFileStore`
- **Changes a knowledge base** `updateKnowledgeBase`
- **Changes a knowledge source** `updateKnowledgeSource`
- **Changes a storage connection** `updateStorageConnection`
- **Changes a workflow webhook** `updatePipelineWebhook`
- **Changes the company's settings for the get file tool** `updateGetFileSettings`
- **Changes the company's settings for the list files tool** `updateListFilesSettings`
- **Changes the company's settings for the save file tool** `updateSaveFileSettings`
- **Changes the skills an agent uses and when it uses them** `updateAgentSkills`
- **Changes which tools a message key may use** `updateMessageKeyToolGrants`
- **Check twilio connection** `checkTwilioConnection`
- **Checks that a storage connection works** `checkStorageConnection`
- **Connect connector preset** `connectConnectorPreset`
- **Create falcon code repository** `createFalconCodeRepository`
- **Create report schedule** `createReportSchedule`
- **Create send grid connection** `createSendGridConnection`
- **Create twilio connection** `createTwilioConnection`
- **Creates a company skill** `createCompanySkill`
- **Creates a file store** `createFileStore`
- **Creates a knowledge base** `createKnowledgeBase`
- **Creates a link to a stored file** `createFileLink`
- **Creates a new company skill and adds it to an agent in one step** `createAndBindAgentSkill`
- **Defines a new company tool** `createCompanyTool`
- **Delete a company skill** `deleteCompanySkill`: Delete a skill from the company library.
- **Delete a knowledge base** `deleteKnowledgeBase`: Delete a knowledge base with its documents and search index.
- **Delete a workflow** `deletePipeline`: Delete a workflow for good.
- **Disconnect remote connector** `disconnectRemoteConnector`
- **Export report** `exportReport`
- **Get agent daily cost cap** `getAgentDailyCostCap`
- **Get agent skill derivation** `getAgentSkillDerivation`
- **Get cost report** `getCostReport`
- **Get report schedules** `getReportSchedules`
- **Get workflow event example** `getPipelineEventExample`
- **Get workflow raise example** `getPipelineRaiseExample`
- **Get workflow report** `getPipelineReport`
- **Get workflow version** `getPipelineVersion`
- **List knowledge base choices** `listKnowledgeBaseChoices`
- **Lists a workflow's webhooks** `listPipelineWebhooks`
- **Lists the company's skills** `listCompanySkills`
- **Lists the company's workflows** `listPipelines`
- **Lists the knowledge bases a connection can reach, so you can add them** `discoverKnowledgeBases`
- **Lists the tools and knowledge searches you can add to an agent** `listToolCatalog`
- **Lists what starts a workflow** `listPipelineTriggers`
- **Makes a copy of a company skill** `duplicateCompanySkill`
- **Moves a company tool through review and release: submit it for review, approve it, publish it or retire it** `actOnTool`
- **Publishes a skill's draft as a new version** `publishCompanySkill`
- **Put tool webhook signing** `putToolWebhookSigning`
- **Put twilio connection** `putTwilioConnection`
- **Reads a skill's published versions and whether its draft is ready to publish** `getCompanySkillPublication`
- **Reads one company skill** `getCompanySkill`
- **Refresh tool catalog** `refreshToolCatalog`
- **Reschedule workflow run** `reschedulePipelineRun`
- **Resolve tool ui action** `resolveToolUiAction`
- **Restores an older workflow version as the draft. To change Live, make an older version live instead** `rollbackPipeline`: Restores an older workflow version as the draft.
- **Retry workflow run** `retryPipelineRun`
- **Runnow workflow run** `runnowPipelineRun`
- **Runs a test call of a company tool. Repeat the same call with the same idempotency key to read its result** `testCompanyTool`: Runs a test call of a company tool.
- **Select connector tools** `selectConnectorTools`
- **Shows the tool a workflow's agents use to report the exceptions the workflow declares** `getPipelineExceptionTool`
- **Starts syncing a knowledge base from its sources** `startKnowledgeSync`
- **Test knowledge search** `testKnowledgeSearch`
- **Test send grid connection** `testSendGridConnection`
- **Test workflow** `testPipeline`
- **Update agent skill derivation** `updateAgentSkillDerivation`
- **Update company messaging rules** `updateCompanyMessagingRules`
- **Update company webhook endpoint** `updateCompanyWebhookEndpoint`
- **Update falcon code repository** `updateFalconCodeRepository`
- **Update report schedule** `updateReportSchedule`
- **Update send grid connection** `updateSendGridConnection`
- **Update step tool settings** `updateStepToolSettings`
- **Update tool implementation choice** `updateToolImplementationChoice`
- **Update tool sign in** `updateToolSignIn`

## Run label index

Make run labels kept before the label index existed findable in a workflow's run search, from the operator console.

- **Index older run labels** `backfillOperatorRunLabelIndex`: Index one company's run labels that were kept before the label index existed, so run search finds them.

## Service callbacks and client transport

Receive authenticated callbacks and exchange conversation and workflow data.

- **Accept a signed guest invitation** `acceptPublicGuestInvitation` (API only): Accept a single-use signed invitation proof without Studio sign-in.
- **Accept a signed tool event** `acceptToolCallEvent` (API only): Message-key grants default empty.
- **Add an agenda item from a company server** `addPersonAgenda` (API only): Contract and synthetic fixture only.
- **Get a reply stream ticket** `createClientStreamTicket` (API only): Exchange the session Bearer, sent in the Authorization header, for a ticket that opens this session's reply stream once.
- **List a person’s agenda** `listPersonAgenda`: Read the agenda items for the referenced person under current conversation-read permission.
- **Prepare a conversation upload** `createClientUpload` (API only): Synthetic client exchange.
- **Raise a signed workflow exception** `raisePipelineException` (API only): Contract and synthetic fixture only.
- **Read a private company theme PNG** `getCompanyThemeAsset`: Read an authorized company theme asset as PNG bytes.
- **Read a voice session** `getVoiceSession` (API only): Read a voice session started with this client session Bearer: whether it is ready, live or ended, why it ended, its length limit and the seconds used so far.
- **Read a workflow run by its server reference** `fetchPipelineRun`: Read a workflow run within the caller’s permitted scope.
- **Read conversation events** `getClientEvents` (API only): Deprecated: the session value in this path can reach load balancer and CDN access logs.
- **Read the reply stream with a ticket** `getClientStream` (API only): Read a client session's replies as server-sent events, opened with a ticket from createClientStreamTicket.
- **Receive a tool completion callback** `receiveToolWebhookCallback` (API only): Bound to tenant, waiting call/version and output schema.
- **Start talking in a chat session** `startVoiceSession` (API only): Start voice in the conversation a client session Bearer names: the public widget, a public link, or a session your own server exchanged for a person with its message key.
- **Stop talking** `endVoiceSession` (API only): End a voice session started with this client session Bearer.
- **Upload conversation content** `putClientUpload` (API only): Synthetic client exchange.

## Storage

Inspect file stores, storage configuration and file-tool examples.

- **List file expiry folders** `listExpiryFolders`: Inspect configured expiry folders and retention rules.
- **List file stores** `listFileStore`: List the company's file stores and their connection assignments.
- **List storage connections** `listStorageConnection`: List configured storage connections and their readiness.
- **Read a file move job** `getFileMoveJob`: Inspect an existing file move job's progress.
- **Read a file store** `getFileStore`: Read a file store's location, settings and revision.
- **Read a file-listing example** `getListFilesExample`: Inspect example file-listing arguments and results.
- **Read a file-reading example** `getReadFileExample`: Inspect example file-reading arguments and results.
- **Read a save-file example** `getSaveFileExample`: Inspect example save-file arguments and results.
- **Read a storage connection** `getStorageConnection`: Inspect a storage connection's configuration and security status.
- **Read a storage policy** `getStoragePolicy`: Read the required storage policy for a selected connection.
- **Read get-file tool settings** `getGetFileSettings`: Inspect the get-file tool's configured access and defaults.
- **Read list-files tool settings** `getListFilesSettings`: Inspect the list-files tool's configured scope and defaults.
- **Read save-file tool settings** `getSaveFileSettings`: Inspect destination and file-handling settings.

## Studio address and custom domain

Each company has one Studio address. Read it, ask to move it to a domain the company owns, prove the domain with a DNS record, and remove the domain again.

- **Ask to move Studio to a custom domain** `setCompanyStudioDomain`: Name a domain the company owns for its Studio and get the TXT record that proves it.
- **Check the custom domain's DNS record** `checkCompanyStudioDomain`: Look for the TXT record now.
- **Read the company's Studio address** `getCompanyStudioAddress`: Read where the company's Studio answers, and any custom domain it is moving to with the DNS records still needed.
- **Remove the custom domain** `removeCompanyStudioDomain`: Drop a custom domain request, or stop using a custom domain in use; the company's Studio then answers at its subdomain again and people sign in there.

## Ask Studio and MCP usage

See what Ask Studio and connected MCP clients cost this month, per person, per client and per operation, and set the company's monthly caps and kill switch.

- **Change Ask Studio and MCP limits** `updateStudioUsageLimits`: Set each person's and each connected client's monthly cap for Ask Studio and MCP calls, or turn the kill switch on or off.
- **Read Ask Studio and MCP usage** `getStudioUsage`: Read a month's Studio assistant and MCP calls lines, with each person, each connected client and the costliest operations, and the caps in force.

## Tools

Inspect tool definitions, safeguards, task progress and integration examples.

- **Ask one question across GitHub files** `askGithubFiles`: Ask one registered question over bounded GitHub file items at an immutable commit.
- **Find relevant github files** `findRelevantGithubFiles`: Rank relevant files in a connected GitHub repository at an immutable commit without returning repository content.
- **List built-in tool definitions** `listBuiltInToolDefinitions`: Inspect the built-in tool kinds and their configuration fields.
- **List company tools** `listCompanyTool`: List tool definitions, safeguards and publication state.
- **List connected code repositories** `listFalconCodeRepository`: List the connected repositories available for code tools.
- **List remote mcp connections** `listRemoteMcpConnections`: List configured remote MCP connections and their status.
- **List tool implementation choices** `listToolImplementationChoices`: Compare the available backend kinds.
- **List tool interface action kinds** `listToolUiActionKinds`: Inspect the supported interface action kinds.
- **Read a background task** `getBackgroundTask`: Read an existing task's progress and result using its task identifier.
- **Read a background task event example** `getBackgroundTaskEventExample`: Inspect a task progress event's shape.
- **Read a company tool** `getCompanyTool`: Read a tool's input and output shapes, backend, safeguards and current revision.
- **Read a connected code repository** `getFalconCodeRepository`: Inspect a connected repository's configuration and state.
- **Read a recorded tool call** `getToolCallRecord`: Read the status and permitted result of an existing tool call using its call identifier.
- **Read a remote MCP connection** `getRemoteMcpConnection`: Read a configured remote MCP connection and its status.
- **Read a tool completion event example** `getToolEventBridgeCompletionExample`: Inspect the completion event format for a tool that finishes asynchronously.
- **Read a tool interface action example** `getToolUiActionExample`: Read an example of a tool result that requests an interface action.
- **Read a tool webhook callback example** `getToolWebhookCallbackExample`: Inspect the callback shape used to complete a deferred tool call.
- **Read a tool webhook request example** `getToolWebhookRequestExample`: Inspect the request shape a tool webhook receives.
- **Read a tool's sign-in requirements** `getToolSignIn`: Inspect the connection and sign-in requirements for a tool.
- **Read agenda tool settings** `getAgendaSettings`: Inspect the agenda tool's configured defaults.
- **Read an agenda event example** `getAgendaEventExample`: Inspect the declared agenda event shape.
- **Read background task limits** `getBackgroundTaskLimits`: Inspect the configured task limits.
- **Read email shared-sender reply settings** `getEmailMessagingSettings`: Read Send email settings, including what to do when a reply could be for more than one open conversation.
- **Read Slack shared-sender reply settings** `getSlackMessagingSettings`: Read Slack sender settings, including what to do when a reply could be for more than one open conversation.
- **Read Telegram shared-sender reply settings** `getTelegramMessagingSettings`: Read Telegram sender settings, including what to do when a reply could be for more than one open conversation.
- **Read text shared-sender reply settings** `getTextMessagingSettings`: Read text sender settings, including what to do when a reply could be for more than one open conversation.
- **Read tool call examples** `getToolCallExamples`: Read example arguments and results to understand a tool's declared interface.
- **Read tool editing options** `getCompanyToolOptions`: Read icon choices and current safeguard authority.
- **Read tool publication state** `getToolPublication`: Inspect the saved tool versions and their publication state.
- **Read tool webhook signing status** `getToolWebhookSigning`: Read webhook signing configuration and key status.
- **Read WhatsApp shared-sender reply settings** `getWhatsappMessagingSettings`: Read WhatsApp sender settings, including what to do when a reply could be for more than one open conversation.
- **Set email shared-sender reply settings** `putEmailMessagingSettings`: Save Send email settings for replies that could be for more than one open conversation.
- **Set Slack shared-sender reply settings** `putSlackMessagingSettings`: Save the Slack sender setting for replies that could be for more than one open conversation.
- **Set Telegram shared-sender reply settings** `putTelegramMessagingSettings`: Save the Telegram sender setting for replies that could be for more than one open conversation.
- **Set text shared-sender reply settings** `putTextMessagingSettings`: Save the text sender setting for replies that could be for more than one open conversation, including the seven-day recent window.
- **Set WhatsApp shared-sender reply settings** `putWhatsappMessagingSettings`: Save the WhatsApp sender setting for replies that could be for more than one open conversation.

## Workflows

Inspect settings saved on workflow steps and the values they inherit.

- **Adopt a Gmail reply draft agent** `adoptGmailReplyDraftTemplate`: Create and publish a Dev reply draft workflow using an existing immutable agent publication containing only the reviewed draft instructions.
- **Create a call-recording-ready workflow starter** `createCallRecordingStarter`: Create a workflow that starts when a call recording is ready, and make it Live.
- **Install the Gmail reply draft workflow** `installGmailReplyDraftTemplate`: Install reviewed draft instructions on an unused root agent, publish its immutable Dev version and create the reply draft workflow.
- **Read effective workflow step tool settings** `getStepEffectiveToolSettings`: Read a step's resolved tool settings and their sources.
- **Read workflow step tool settings** `getStepToolSettings`: Read the settings saved directly on a workflow step.
- **Receive a finished call recording** `receiveCallRecording` (API only): Receive a provider-signed finished recording notification for the transcript and report starter.