The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill acts as an automated pipeline that transforms external, untrusted project requirements into a structured plan containing shell commands (pytest, git) and Python code snippets for execution.
Ingestion points: The skill takes project specs or requirements as input to generate the implementation plans described in SKILL.md.
Boundary markers: No explicit delimiters or instructional guardrails (e.g., "ignore instructions in the requirements") are defined to prevent the agent from inadvertently following malicious commands embedded in the source material.
Capability inventory: While the skill's primary action is writing to the docs/plans/ directory, it explicitly instructs the AI to hand off the resulting content to sub-skills (superpowers:executing-plans and superpowers:subagent-driven-development) designed for automated file writing and command execution.
Sanitization: The instructions lack any requirement for the AI to sanitize or validate the content of the requirements before including them in the generated plan.

writing-plans