The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external task specifications or requirement documents to generate multi-step implementation plans.
Ingestion points: User-provided specifications or requirement documents used to generate plans in SKILL.md.
Boundary markers: The skill uses structured templates and specific file paths for output (docs/superpowers/plans/), though it lacks explicit delimiters for the input specifications themselves.
Capability inventory: The generated plans involve file system modifications and execution of testing/git commands as defined in the task templates.
Sanitization: The skill includes a robust "Self-Review" checklist and a mandatory human handoff phase ("Execution Handoff") where a user must approve the plan before it is executed by other sub-skills, serving as a primary defense against malicious instructions embedded in input specs.
[COMMAND_EXECUTION]: The skill instructs the agent to generate plans containing shell commands for standard development workflows.
Evidence: Templates in SKILL.md include pytest for verification and git commands (git add, git commit) for version control. These commands are localized to the project environment and are standard for the skill's intended purpose of software implementation planning.

writing-plans