What problem does it solve? Working on the Anakot Agent Docker image's s6-overlay supervision tree is error-prone: services fail to start, per-profile gateways crash-loop after docker restart, and the container exit code behaves unexpectedly. This Skill provides the architectural context, file map, and operational recipes needed to safely modify, debug, or extend the supervision setup. ## Core Features & Use Cases - Architecture Reference: Explains the Architecture B pattern where the container CMD (main-wrapper.sh) runs as the main program under /init, why s6-supervised main was abandoned, and how cont-init.d, s6-rc.d, and /run/service interact. - Service Management Recipes: Step-by-step instructions for adding new static s6-rc services, changing per-profile gateway run scripts via S6ServiceManager._render_run_script, and manually controlling services with s6-svc. - Troubleshooting Playbook: Diagnoses common failures such as command not found via docker exec, root-owned profile directories breaking the reconciler, crash-looping gateways, and containers exiting 143. - Use Case: A developer adds a new supervised service to the Docker image: create the s6-rc.d/<name> directory with type/run files, wire it into the user bundle, and verify with the docker test harness expecting 19 passed tests. ## Quick Start Ask the agent to explain why a per-profile gateway is not surviving docker restart and show how to inspect its s6 service state.