agent-autonomy-kit
par ryancampbell
AI & LLMsRésumé
Stop waiting for prompts.
Description
Ce que fait ce skill
A framework for running agents proactively using a markdown-based task queue and heartbeat loop. Agents pull tasks from QUEUE.md, execute them, and update status without waiting for user prompts. Cron jobs handle overnight operation and daily reporting.
Eliminates the need to manually re-prompt an agent after each task, so work continues until the queue is empty or resource limits are hit.
Quand l'utiliser
- Running a coding task backlog while away from the keyboard
- Generating daily progress reports automatically at end of day
- Processing a queue of research tasks overnight
- Keeping an agent working across multiple sessions without re-prompting
- Delegating a multi-step project plan to an agent to complete independently
Exemple de workflow
User asks: 'Work through my task queue while I'm away'
- 1
Reads QUEUE.md and identifies all tasks in the Ready section
- 2
Moves the first Ready task to In Progress and updates HEARTBEAT.md
- 3
Executes the task autonomously
- 4
Marks the task Done in QUEUE.md and pulls the next Ready item
- 5
Generates a daily summary report when the queue is empty or limits are reached
Full task queue processed with a written report of completed work
Installation
clawhub install ryancampbell/agent-autonomy-kitExécute la commande CLI, ou colle l'instruction dans Claude Code, Cursor, Codex ou l'agent de code que tu utilises.