The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted user specifications to generate implementation plans and shell commands.
Ingestion points: The skill ingests user-provided specifications and requirements as defined in the description and Overview section in SKILL.md.
Boundary markers: No delimiters or specific instructions are provided to the agent to distinguish between the skill's logic and potentially malicious instructions within the input data.
Capability inventory: The skill writes Markdown files to the filesystem (docs/plans/) and encourages the use of sub-skills for task execution (e.g., superpowers:executing-plans and superpowers:subagent-driven-development) which may execute the generated commands.
Sanitization: No sanitization, filtering, or validation of the input specifications is performed before they are incorporated into the plan document.

writing-plans