The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or unauthorized data access attempts were detected. The skill's behavior aligns with its stated purpose of drafting compliance documentation.
[PROMPT_INJECTION]: The skill exhibits a potential surface for indirect prompt injection as it processes untrusted data from the web and project files.
Ingestion points: Reads project documents and external regulatory guidance using Read, WebSearch, and WebFetch tools (Step 1).
Boundary markers: None specified to distinguish between instructions and ingested data.
Capability inventory: Includes Bash for command execution and Write for file modification.
Sanitization: The instructions do not include specific steps to sanitize or ignore instructions embedded in the ingested content. This is a common risk for context-aware agents and is considered a low-severity architectural concern.

bind-policy