Iris

Iris docs

Iris records what you do on screen and turns it into structured context your AI agents (and you) can read. These docs cover every way to capture and consume that data.

If you have 5 minutes

  1. Create an account.
  2. Hit Record on the dashboard, do something, hit stop.
  3. Open Integrations, generate an API key, and paste the MCP config into Claude Desktop.

Capture surfaces

  • Browser extension — Chrome and Firefox. Cross-tab recording with click/type/navigation tracking.
  • CLI — upload a screen recording from your terminal, list, search, fetch context.
  • TypeScript SDK — call the API from any Node app.

Use the data

  • MCP for Claude — give Claude Desktop read-only access to your recordings as tools.
  • REST API — direct HTTP access for any custom agent or integration.

Operate

  • Self-hosting — Docker image, required env vars, and storage notes.
  • FAQ — quotas, plans, data handling.

Found a gap or a typo? Open an issue or PR on GitHub.