Autonomous Skill
Manages and executes complex, long-running tasks through automated decomposition, multi-session tracking, and headless mode auto-continuation.
Streamlines the execution of time-intensive projects by implementing a dual-agent pattern (Initializer and Executor) that allows Claude to work autonomously across multiple sessions. The skill automatically generates task names, creates persistent state directories for progress tracking, and utilizes a headless execution loop to complete complex objectives like codebase refactoring or full-feature implementation without manual intervention. It ensures task isolation and provides detailed session-by-session reporting, making it perfect for 'vibe coding' and heavy-duty developer workflows.
主な機能
01Comprehensive session reporting and progress summarization
02Automated task decomposition and persistent tracking via task_list.md
03Isolated directory structure to support parallel autonomous tasks
04Headless auto-continuation loop with configurable session delays
05925 GitHub stars
06Dual-agent architecture utilizing Initializer and Executor modes
ユースケース
01Executing multi-step maintenance tasks that exceed standard LLM context or time limits
02Building complete REST APIs or complex features from a single natural language description
03Refactoring large modules or migrating authentication systems across multiple sessions