The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes external specifications and requirements to generate executable implementation plans. These plans are intended to be executed via sub-skills like superpowers:executing-plans. A malicious specification could potentially inject harmful commands or code into the generated plan, which the agent might then execute.
Ingestion points: External specifications and requirements provided by the user or fetched from the environment.
Boundary markers: No explicit markers are defined to isolate untrusted specification content from the generation logic.
Capability inventory: The skill interacts with superpowers:executing-plans and superpowers:subagent-driven-development, both of which have capabilities for shell command execution and file system modification.
Sanitization: There is no mention of sanitization, filtering, or validation of the input specification content before it is used to generate the plan.

writing-plans