The Agent Skills Directory

[SAFE]: The skill's functionality is limited to managing internal project documentation and state. No unauthorized network access, sensitive data exposure, or malicious code execution patterns were found.\n- [PROMPT_INJECTION]: The skill reads data from potentially untrusted sources like REVIEW.md and feature artifacts to generate learning records. While this constitutes an indirect prompt injection surface, it is a core part of the intended functionality and operates within restricted project paths.\n
Ingestion points: REVIEW.md and feature-local learning records.\n
Boundary markers: Absent; the agent is instructed to process content directly.\n
Capability inventory: Writes to .beads/learnings/ and specific shared state fields; uses beads-cli for provenance.\n
Sanitization: No explicit sanitization of ingested text is performed.

beo-compound