orbibuild/orbi repository overview

⁠Orbi Docker image

Orbi⁠ is an AI coding agent that runs on GitHub Issues: label an Issue ai-ready and it implements, tests, reviews in a separate session, merges and tags. This image is a community-maintained wrapper around the unmodified official deployment: systemd as PID 1 running the same units the systemd install uses.

Tags: latest and the release number without the v (e.g. 0.5.18), for linux/amd64 and linux/arm64. The same image is on GHCR as ghcr.io/orbi-build/orbi.

One docker run reaches a real delivery. Substitute the three quoted values: a GitHub token with write access to the task-pool repo, the repo as OWNER/REPO, and the model key. DeepSeek is the example; any OpenAI-compatible endpoint works through the same four ORBI_PI_* variables.

docker run -d --name orbi \
  --stop-signal SIGRTMIN+3 \
  --tmpfs /run --tmpfs /tmp \
  --cgroupns=host -v /sys/fs/cgroup:/sys/fs/cgroup:rw \
  -v orbi-deploy:/orbi \
  -v orbi-work:/work \
  -e GH_TOKEN="github_pat_xxx" \
  -e ORBI_SOURCE_REPO="OWNER/REPO" \
  -e ORBI_PI_PROVIDER=deepseek \
  -e ORBI_PI_MODEL=deepseek-chat \
  -e ORBI_PI_BASE_URL=https://api.deepseek.com \
  -e ORBI_PI_API_KEY="sk-xxx" \
  docker.io/orbibuild/orbi:latest

Then open an Issue in OWNER/REPO and add the ai-ready label. Within two 5-minute ticks a PR opens; it merges after the review round. All state lives in the two named volumes.

• Orbi⁠: what it is and why, self-hosted or cloud
• Full Docker guide⁠: every variable, the two volumes, the token permissions, and the Orbi quick guide (Issue shape, labels, releases)
• Source of this image⁠
• Orbi repository⁠