piflow-skill-generator_planner

Run scripts/generate_piflow_skill.py with --spec pointing to a UTF-8 JSON spec containing name, description, input_params, output_params, and a script definition. The script creates the full skill directory under skills/generated and registers it.

Pass a flow summary JSON via --flow to generate_piflow_skill.py. The generator restores a draft spec from the flow's inputs, outputs, scripts, and processing steps, then builds the skill directory using the standard template.

Each skill needs a SKILL.md with YAML frontmatter (name, description, version, input_params, output_params), a skill.json with role-annotated parameters and command_template, and an executable Python entry script under scripts/. References and assets directories are optional.

Yes, use scripts/rewrite_piflow_skill.py with --skill-dir and a new flow summary. It merges existing stable facts with the new flow, keeps the original skill name, and rebuilds the directory with overwrite enabled.

Common causes include mismatched command_template tokens versus input_params, non-UTF-8 encoding, directory name differing from the frontmatter name, or missing required parameter fields. Run validate_piflow_skill.py to identify the specific issue.

It should only run when the user explicitly asks to generate or save a skill, or when a completed task's verified workflow needs to be persisted. It is not a default entry point for general data processing requests.