discussion-participation
Pass
Audited by Gen Agent Trust Hub on Nov 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill identifies and mitigates the risk of indirect prompt injection from untrusted GitHub Discussion content. • Ingestion points: Fetches thread data via get_discussion and get_discussion_comments tools. • Boundary markers: Instructions explicitly mandate treating comments as untrusted data and ignoring embedded instructions. • Capability inventory: The agent can read, write, update, and delete comments. • Sanitization: Behavioral constraints are used to ensure the agent does not follow instructions found within discussion threads.
- [NO_CODE]: The skill contains no executable scripts or external code dependencies, consisting only of instructional markdown.
Audit Metadata