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

# project.md

> project.md is the user-managed memory file Qodex reads on every agent turn.

# project.md

`project.md` is the main project memory file you control. Qodex reads it on every agent turn so the agent understands your app before it starts testing.

Use it for stable context: what the app does, how auth works, which accounts exist, what to focus on, and what to avoid.

## What belongs in project.md

| Section          | What to include                                                       |
| ---------------- | --------------------------------------------------------------------- |
| App overview     | What the product does and who uses it.                                |
| Stack            | Frameworks, database, hosting, important services.                    |
| Test accounts    | Emails and roles. Put passwords in environment variables, not memory. |
| Known quirks     | Expected behavior that might look broken.                             |
| Focus areas      | Product flows Qodex should test carefully.                            |
| Skip areas       | Routes or actions Qodex should avoid.                                 |
| Team conventions | Rules the agent should respect.                                       |

## Example structure

```markdown theme={null}
# Project

## App overview

Qodex is an autonomous AI QA platform. Customers run it against their APIs
and web apps to find bugs before they ship.

## Stack

- Frontend: React + Vite
- Backend: Node.js + Hono
- Database: Postgres

## Test accounts

- Admin: admin@example.com, password from AUTH_PASSWORD
- Member: member@example.com
- Viewer: viewer@example.com

## Known quirks

- The dashboard can take a few seconds to warm up after deploy.
- /admin/reset is destructive. Never test it outside local dev.

## Focus areas

- PR review walkthroughs.
- Scenario authoring across API and UI.
```

## What does not belong

Keep these out of `project.md`:

* Passwords, API keys, tokens, or secrets.
* PII or customer data.
* Long chat transcripts.
* Per-scenario setup steps.
* Large architecture documents that belong in a wiki.

## How updates work

Edit `project.md` from the Knowledge tab. Changes apply on the next agent turn. There is no rebuild step.

Sub-agents get the same `project.md` context as the main agent, so these facts apply across scenario authoring, review, and verification.

## Related

<CardGroup cols={2}>
  <Card title="Editing and curating memory" icon="pencil" href="/memory-editing-and-curating">
    Keep memory small and accurate.
  </Card>

  <Card title="How memory works" icon="arrow-right-to-bracket" href="/kb-how-memory-works">
    Understand prompt loading and on-demand files.
  </Card>

  <Card title="Projects" icon="folder" href="/concepts-projects">
    See how project boundaries scope memory.
  </Card>

  <Card title="Skills" icon="puzzle" href="/concepts-skills">
    Learn how skills use memory while authoring tests.
  </Card>
</CardGroup>