ci-fix
Pass
Audited by Gen Agent Trust Hub on Feb 22, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [Prompt Injection] (LOW): Risk of indirect prompt injection through log analysis.
- Ingestion points: Reads GitHub Action logs via
gh run view --log-failedandgh run view --log --job <job-id>. It also downloads artifacts viagh run download. - Boundary markers: No delimiters or instructions are provided to the agent to treat log or artifact content as untrusted data.
- Capability inventory: The agent has the ability to write code, perform
git commit, andgit pushto a new branch. - Sanitization: No sanitization or filtering of external content is performed before the agent analyzes it to determine the root cause and implement a fix.
- [Command Execution] (SAFE): The skill uses
gh(GitHub CLI) andgitcommands which are necessary for its primary purpose of diagnosing and fixing CI issues.
Audit Metadata