lab42it/loopq repository overview

⁠loopq

Your coding agents, all the loops you want: assign, review, integrate. Locally.

loopq coordinates coding agents in separate Git worktrees. It assigns tasks, collects results, gets a second agent identity to review them, and integrates approved changes into a dedicated base branch. Each project loop has its own agents, worktrees, and queue. Everything runs on your machine, under your control.

⁠Tags

• latest: most recent release
• X.Y.Z: a specific release, for example 0.3.1
• X.Y: latest patch of a minor line, for example 0.3

Images are multi-arch (linux/amd64, linux/arm64) and also published to ghcr.io/alvagante/loopq.

⁠Usage

Mount your loop configs (read-only) and the queue data, and run as your own user so files stay owned by you:

docker run --rm --user "$(id -u):$(id -g)" \
  --mount "type=bind,source=$HOME/.config/loopq/loops.d,target=/config,readonly" \
  --mount "type=bind,source=$HOME/.loops,target=/data" \
  -e LOOPQ_CONFIG_DIR=/config \
  lab42it/loopq status --all

The entrypoint is loopq, so any subcommand works (status, list, runs, stats, doctor, ...). With no arguments it runs doctor.

⁠Notes

• LOOPQ_HOME is /data (a volume); the working directory is /work.
• The image includes Git and the loopq dependencies, but not your agent binaries. Commands that launch agents need those binaries, plus access to the configured worktrees, inside the container.
• Most people install the script directly with ./manage.sh install; the image is mainly for inspecting loops and running ticks in containers.

⁠Links

• Source and docs: https://github.com/alvagante/loopq⁠
• Changelog: https://github.com/alvagante/loopq/blob/main/CHANGELOG.md⁠
• License: Apache-2.0