Build a custom Runloop devbox image with a Blueprint (Dockerfile or launch commands)

Documented steps

1. Blueprints are reusable, cacheable startup templates; build from a Dockerfile via CLI: `rli blueprint from-dockerfile --name my-agent-env --dockerfile ./docker/Dockerfile --build-context .`
2. Or create programmatically: Python `blueprint = await runloop.blueprint.create(name="docs-template", launch_parameters={"launch_commands": ["sudo apt install -y jq"]})`; TypeScript `runloop.blueprint.create({ name: ..., launch_parameters: {...} })`.
3. Spin up a devbox from the blueprint: `devbox = await blueprint.create_devbox()` (Python) / `blueprint.createDevbox()` (TypeScript).
4. Check build status: `blueprint = await runloop.blueprint.from_id(blueprint.id)` then get_info() — status values include queued, build_complete, build_failed (build queue max length is 2000).
5. Retrieve build logs: `await blueprint.logs()` (SDK) or `rli blueprint logs <blueprint_id>` (CLI). Build logs are retained for 28 days after completion.
6. Any background/long-running command in launch_commands must redirect its output (e.g. `"my-server start > /tmp/server.log 2>&1 &"`) or the devbox will never reach an active state.
7. Docs: https://docs.runloop.ai/docs/devboxes/blueprints/overview