feature-run

Batch processing automates running all pending features in sequence through a complete workflow loop. This Skill processes your entire feature backlog continuously, making autonomous decisions and handling errors without requiring human intervention or questions, so progress continues unattended until all tasks complete.

Yes, single feature mode lets you focus on and complete one specific task by providing its task ID. The Skill applies the same autonomous workflow—next, implement, check, done—to that individual feature, handling errors and verification independently.

Unattended mode handles failures autonomously without pausing or asking for input. The Skill continues processing remaining tasks, logs structured results including task_id, status, and verification outcome, and completes the entire workflow cycle regardless of individual task failures.

The Skill outputs structured result parsing for each task: task_id, status (pass or fail), verification_passed (boolean), files_modified (list), and notes (summary). You review the final report to see which features succeeded, what code changed, and whether verification passed.

The Skill enforces a strict four-step sequence: next (identify task), implement (write code), check (verify results), done (mark complete). This CLI-only workflow ensures consistent progress through your feature backlog with no deviation or manual approval gates.

Yes, unattended batch processing is designed exactly for this use case. Define your project goals and features, then run the Skill to let the AI agent work through your entire backlog autonomously, handling decisions and issues so you focus on other priorities.