skills/phrazzld/claude-config/fix-ci/Gen Agent Trust Hub

fix-ci

Pass

Audited by Gen Agent Trust Hub on Feb 20, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION] (SAFE): The skill utilizes the GitHub CLI (gh) for routine operations such as listing runs (gh run list) and viewing logs (gh run view). These are standard administrative tasks for the intended use case.
  • [INDIRECT_PROMPT_INJECTION] (LOW): The skill processes external data (CI logs) via gh run view --log. This presents a theoretical surface where malicious logs could attempt to influence the agent's analysis.
  • Ingestion points: Output from gh run view <run-id> --log.
  • Boundary markers: None specified for log content.
  • Capability inventory: File creation (markdown summaries) and PR updates.
  • Sanitization: None specified; the agent is expected to parse raw log text.
  • [DATA_EXPOSURE] (SAFE): While CI logs can occasionally contain sensitive information, the skill does not exfiltrate this data or perform unauthorized network requests. It follows standard DevOps practices for log inspection.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 20, 2026, 11:43 PM