skills/shipshitdev/skills/bug/Gen Agent Trust Hub

bug

Pass

Audited by Gen Agent Trust Hub on Sep 6, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-provided bug descriptions and error logs to generate issues.\n
  • Ingestion points: User bug descriptions provided as free text or stack traces processed in Phase 2 of SKILL.md.\n
  • Boundary markers: Employs a mandatory Phase 3 preview and confirmation step where the agent must wait for an explicit user 'yes' before executing any creation commands.\n
  • Capability inventory: Utilizes gh issue create, gh label create, gh api, and git commands to manage repository metadata and issue creation.\n
  • Sanitization: Implements high-quality sanitization by writing the structured report to a temporary local file (.tmp/bug_body.md) and using the GitHub CLI's --body-file flag, which effectively prevents shell command injection from malicious characters in the bug description.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 6, 2026, 08:30 AM
Security Audit — agent-trust-hub — bug