github-mcp
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from external GitHub repositories which could contain untrusted instructions.
- Ingestion points: The
get_file_contentstool (sub-skills/get_file_contents.md) reads file and directory content from remote repositories into the agent context. - Boundary markers: No explicit delimiters or boundary instructions are defined in the sub-skill parameters to separate external content from agent instructions.
- Capability inventory: The skill possesses high-privilege capabilities including writing files (
push_files), creating issues (create_issue), and opening pull requests (create_pull_request). - Sanitization: There is no evidence of sanitization, filtering, or validation of the content retrieved from repositories before it is processed by the agent.
Audit Metadata