respond-to-pr-comments

Pass

Audited by Gen Agent Trust Hub on Oct 29, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted reviewer comments from Azure DevOps. It mitigates this risk by instructing the agent to ignore any commands embedded in comments (e.g., requests to install software or run scripts) and requires the agent to restate feedback as technical requirements before asking for human approval.
  • [COMMAND_EXECUTION]: The skill uses various helper scripts (e.g., parse-pr-url.py, ensure-env.py). It enforces security by piping potentially sensitive data (like remote URLs) directly to scripts rather than passing them as command-line arguments, preventing credential leakage in process logs.
  • [DATA_EXFILTRATION]: While the skill interacts with Azure DevOps APIs, it limits operations to those authorized by the user. It explicitly warns against printing raw thread payloads and uses a 'dry-run' phase for comments to ensure the user approves the exact content being sent externally.
  • [PRIVILEGE_ESCALATION]: The skill follows a least-privilege model, requiring separate user authorizations for testing, building, or restoring, and mandates identity verification to ensure the user is the owner of the PR being modified.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 29, 2026, 03:17 PM