triage

Pass

Audited by Gen Agent Trust Hub on Oct 26, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external content (GitHub issues and pull requests) and perform high-privilege actions like running shell commands and writing to the repository.
  • Ingestion points: The triage workflow in SKILL.md reads the full body, comments, and diffs of issues and external pull requests.
  • Boundary markers: There are no explicit instructions or delimiters used to separate the untrusted external content from the agent's internal instructions, nor are there warnings to ignore embedded commands within the ingested data.
  • Capability inventory: The agent has the capability to execute shell commands (check it out, run the relevant tests or commands), write to the file system (update GLOSSARY.md/ADRs, write to .out-of-scope/), and interact with the issue tracker (post comments and close issues).
  • Sanitization: The skill does not prescribe any sanitization, validation, or human review process for the commands or tests identified within the untrusted external content before they are executed.
  • [COMMAND_EXECUTION]: The skill explicitly instructs the agent to execute commands from the codebase to verify claims.
  • Evidence: In SKILL.md, step 3 (Verify the claim) directs the agent to "check it out, run the relevant tests or commands." This applies to external pull requests, meaning the agent may execute code provided by untrusted contributors.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 26, 2026, 09:28 AM