> ## Documentation Index
> Fetch the complete documentation index at: https://familyco.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Chat

> The Founder's command center — directives, sessions, and structured replies.

Chat is the primary entry point for the Founder. Each conversation is a session bound to an agent; directives become AgentRuns that touch projects, tasks, and the inbox.

## UI areas

* **Conversation list** — sessions per executive agent.
* **Message timeline** — Founder messages and agent replies.
* **Context sidebar** — referenced projects, tasks, agents, and inbox items.
* **Quick actions** — open referenced entities, approve from inbox, jump to audit.

## Behaviors

* The Founder can send a directive at any time.
* Multiple chat sessions per agent are supported; switch via the session sidebar.
* The most recently selected session is restored when the Founder returns to Chat.
* Messages may reference projects, tasks, agents, and inbox items.
* AI replies use a structured format: **understanding**, **plan**, **approvals**, **next steps**.
* Chat displays when an action created or updated entities and links to them.

## How a directive flows

1. Founder sends a message.
2. Server creates an `AgentRun` with `triggerType = founder_chat`.
3. Executive Agent interprets, plans, and may create or update Projects/Tasks.
4. If the action requires approval, the run pauses (`waiting_approval`) and an Inbox item is created.
5. The Founder approves, rejects, or answers — the run resumes from its checkpoint.
6. Outcomes return as a structured chat reply with linked entities.

## Tool context per session

Each session receives a tool list filtered by the agent's level (`L0`/`L1`/`L2`) plus any plugin tools the company has enabled. Built-in tools are immutable; plugin tools require valid custom field values before they become available.

See [Tools](/modules/tools) for the policy model and [Plugins](/modules/plugins) for capability registration.
