The Agent Skills Directory

[PROMPT_INJECTION] (LOW): Indirect Prompt Injection surface detected in documentation workflows.
Ingestion points: The skill ingests untrusted data from the docs/ directory using the Read and Grep tools during the deduplication and staleness detection phases in SKILL.md.
Boundary markers: Absent. There are no instructions to the agent to treat the contents of the documentation files as data rather than instructions.
Capability inventory: The skill possesses Write and Edit capabilities, which could be misused if a processed document successfully injects instructions into the agent's context.
Sanitization: Absent. No sanitization or validation is applied to the content of the markdown files before they are read or edited by the agent.

knowledge-capture