github-api

Pass

Audited by Gen Agent Trust Hub on May 14, 2026

Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill communicates with non-whitelisted domains including gateway.maton.ai, ctrl.maton.ai, and connect.maton.ai to proxy GitHub requests. It transmits the MATON_API_KEY environment variable to these endpoints as part of its core functionality.\n- [PROMPT_INJECTION]: The skill possesses a significant attack surface for indirect prompt injection due to its broad read and write capabilities on GitHub repositories.\n
  • Ingestion points: Untrusted data is ingested into the agent context via multiple endpoints, specifically GET /github/repos/{owner}/{repo}/contents/{path}, GET /github/repos/{owner}/{repo}/issues, and GET /github/repos/{owner}/{repo}/pulls.\n
  • Boundary markers: The instructions do not define delimiters or provide clear guidance to the agent to distinguish between trusted instructions and retrieved repository content.\n
  • Capability inventory: The skill provides the agent with powerful capabilities, such as repository deletion (DELETE /github/repos/{owner}/{repo}), file creation/modification (PUT /github/repos/{owner}/{repo}/contents/{path}), and pull request merging, which could be maliciously exploited if the agent obeys instructions embedded in external data.\n
  • Sanitization: There is no evidence of data validation, sanitization, or filtering applied to the content retrieved from GitHub before it is processed.
Audit Metadata
Risk Level
SAFE
Analyzed
May 14, 2026, 12:15 AM
Security Audit — agent-trust-hub — github-api