Pi Workflow Orchestration

Workflow orchestration for Pi's task management, self-improvement, and code quality standards. Use when starting new projects, managing multi-step tasks (3+...

This skill appears to do exactly what it says: provide planning templates, capture lessons, and optionally inject a session-start reminder. Before enabling/installing: 1) Review handler.js/handler.ts (they are short and only inject a virtual file) and confirm you are comfortable copying the hook into ~/.openclaw/hooks; 2) If you plan to run scripts/sync_lessons.py, use the --dry-run first to confirm it reads the correct workspace path (~/.openclaw/workspace by default) and that you’re ok with it updating references/lessons.md in the skill repo; 3) Be cautious if you edit the handler to add links — pointing to internal URLs is fine but avoid adding network calls or secret embeddings; 4) If you prefer no automatic reminders, keep the hook disabled (openclaw hooks enable/disable is documented). Overall the files are coherent with the stated purpose and do not request unrelated privileges or credentials.

Type: OpenClaw Skill Name: pi-workflow Version: 1.1.1 The skill bundle provides a structured workflow for an AI agent's self-improvement and task management. It instructs the agent to write to specific markdown files within its workspace (`tasks/lessons.md`, `tasks/errors.md`, `tasks/feature_requests.md`). A Python script (`scripts/sync_lessons.py`) handles syncing these workspace lessons to the skill's `references/lessons.md` file, performing only local file read/write operations within the expected scope. An OpenClaw hook (`hooks/openclaw/handler.js`/`.ts`) injects a virtual markdown reminder (`PI_WORKFLOW_REMINDER.md`) into the agent's context at session start, which is a benign form of self-guidance. All instructions and code are aligned with the stated purpose and show no evidence of malicious intent, data exfiltration, unauthorized execution, or persistence mechanisms.