The Agent Skills Directory

[PROMPT_INJECTION]: The skill demonstrates an indirect prompt injection attack surface as it is designed to ingest and analyze user-provided documentation files without explicit boundary markers.
Ingestion points: Discovery Phase involves reading all documentation files related to a topic (SKILL.md).
Boundary markers: Absent; there are no instructions to the agent to disregard instructions found within the processed text.
Capability inventory: The skill has the capability to delete files and modify project structure files like CLAUDE.md and README (SKILL.md).
Sanitization: Absent; content is analyzed for value but not filtered for instructional patterns.
[COMMAND_EXECUTION]: The execution phase of the skill workflow explicitly directs the agent to delete source files identified as redundant. This represents a high-impact file system operation associated with the skill's primary purpose.
[PROMPT_INJECTION]: The presence of the .security-scan-passed file constitutes a self-referential safety claim. These claims attempt to influence the analysis of the skill's safety and are treated as data to be evaluated rather than authoritative verdicts.

docs-cleaner