Agent GTD

This skill is internally consistent with its stated goal of local GTD-style agent management. Before installing or running it: - Review and run the helper scripts (scripts/*.py, scripts/sleep.sh) manually to see what they do. They call 'task' and 'timew' and read/write local files (e.g., ~/.agent/last_active, ops/session_state.md). - The runbook references an automated promotion script (promote_learnings.py) but that file is not present in the bundle; if you plan to rely on automated cron jobs, confirm the implementation exists and inspect it. - If you enable any optional integrations (Pueue bridge, ClawVault, or any Taskwarrior hooks that execute pueue_cmd), audit those components carefully because they can cause background execution of arbitrary commands. The current bundle does not include an auto-execution bridge, but the documentation describes one as a future/adjacent component. - Run the vitality_check script in a safe environment first; it invokes 'timew export' and 'task active', and its fallback reads ~/.agent/last_active — ensure those paths are acceptable for your setup. - Consider running this skill in a sandboxed user account or VM until you are comfortable, and ensure your Taskwarrior data is backed up before bulk changes. If you want narrower assurance, provide the missing promote_learnings.py (if intended) and confirm whether you plan to enable any auto-execution bridges; with that info I can raise confidence to high or flag additional concerns.