backlog-triage

Pass

Audited by Gen Agent Trust Hub on Sep 30, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from GitHub Issue titles, bodies, and comments to generate an advisory report. This report contains machine-readable 'anchor' comments that define GitHub mutations.
  • Ingestion points: Content is fetched from GitHub using gh issue list and gh issue view commands as described in SKILL.md and implemented in scripts/triage-apply.js.
  • Boundary markers: The skill requires a human to manually check boxes ([ ] to [x]) next to specific proposals in the report. The scripts/triage-apply.js script only executes actions where the paired checkbox is marked as checked.
  • Capability inventory: The skill possesses the capability to comment on, close, and edit (labels/milestones) GitHub issues using the gh CLI via execFileSync calls in scripts/triage-apply.js.
  • Sanitization: The scripts/triage-apply.js utility uses quoteShellArg to escape CLI parameters, mitigating basic shell injection risks. The primary risk remains the potential for the agent to include malicious instructions from issue content as valid proposals, though this is mitigated by the manual review step.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 30, 2026, 12:27 AM