The Agent Skills Directory

[SAFE]: The skill provides structured templates and instructions for writing development plans. It uses standard software engineering commands such as pytest and git in an illustrative context for documentation purposes. No network access or high-privilege operations are requested or performed by the skill itself.- [INDIRECT_PROMPT_INJECTION]: The skill contains an attack surface for indirect prompt injection as it processes project requirements to create implementation tasks.
Ingestion points: The skill reads user-provided specifications and requirements as noted in SKILL.md.
Boundary markers: Task decomposition is enforced using a mandatory markdown header and specific checkbox-based step structures to maintain focus.
Capability inventory: The skill is limited to writing markdown files to the docs/superpowers/plans/ directory; it does not possess direct command execution or network capabilities.
Sanitization: The included plan-document-reviewer-prompt.md serves as a validation layer by using a secondary agent to verify that the generated plan aligns with the spec and lacks placeholder content or contradictory steps.

writing-plans