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

# Core Concepts

> Understand how Organizations, Projects, Sessions, and Context Files work together in Brunel Agent.

Brunel is organized around four levels: **Organizations**, **Projects**, **Sessions**, and **Context Files**. Understanding how these relate to each other will help you get the most out of the platform.

***

## Organizations

An Organization is the top-level container in Brunel — it represents your company, team, or any group of people collaborating together. Everything in Brunel lives inside an organization: projects, sessions, members, billing, and settings.

**Key things to know:**

* When you sign up, you'll be prompted to create your first organization
* A single Brunel account can belong to **multiple organizations** — useful if you work across multiple teams or clients. On every login, you'll see the organization selection screen and choose which one to enter
* Organizations have their own member roster, roles, and billing subscription
* If you've been invited to join an existing organization, accept the invitation from your email — you'll be placed directly into that org without needing to create one

***

## Projects

A Project represents a codebase, product, or initiative. Projects live inside an organization and are the container for your planning sessions.

**Key things to know:**

* Create one project per codebase or distinct area of work — for example, `backend-api`, `mobile-app`, or `platform-infra`
* All members of an organization can see all projects, subject to their role permissions
* Projects can be edited (renamed, updated description) or archived. Archived projects are recoverable — they're hidden from the active view but nothing is deleted

***

## Sessions

A Session is a single AI planning conversation within a project. Each session has a type, a name, a full message history, and a lifecycle phase.

**Key things to know:**

* Sessions are the core unit of work in Brunel — create one per task or ticket
* All session history is persistent and cloud-synced. Closing the app, switching tabs, or switching team members does not lose any messages
* Multiple sessions can be open simultaneously in the tab interface

### Session Types

| Type                 | Use when...                                           |
| -------------------- | ----------------------------------------------------- |
| **Feature Planning** | Defining requirements for something new               |
| **Bug Fix**          | Diagnosing an issue and planning a targeted fix       |
| **Refactoring**      | Improving code without changing external behavior     |
| **General Planning** | Spikes, research, migrations, or anything exploratory |

### Session Lifecycle

Every session moves through four phases — all transitions are manual:

```
Backlog → Planning → Execution → Verification
```

| Phase            | Meaning                                            |
| ---------------- | -------------------------------------------------- |
| **Backlog**      | Created but not yet actively being planned         |
| **Planning**     | Actively building the plan with AI                 |
| **Execution**    | Plan has been exported; coding agent is working    |
| **Verification** | Agent work is complete; verifying against the plan |

***

## Context Files

Context Files are documents attached to a session that give Brunel's AI the background it needs to plan accurately for your specific codebase and team.

**Key things to know:**

* Supported formats: plain text, Markdown, code files, `.docx`, `.xlsx`
* Files are stored in the cloud and associated with the session they were uploaded to
* Docx and xlsx files have inline viewers — no need to download them to read them
* The right context files dramatically improve plan quality — see [Context Files](/planning/context-files) for guidance on what to include

***

## MCP Integration

Brunel runs a local MCP (Model Context Protocol) server that coding agents can connect to directly. Once connected, your agent can read plans, context files, and session metadata without any copy-pasting.

See [MCP Integration](/planning/mcp-integration) for full setup instructions.