cmd-fix-github-actions
Pass
Audited by Gen Agent Trust Hub on Apr 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the GitHub CLI (gh) and Git to fetch workflow statuses, view logs, and manage branches. It also executes local verification commands like 'mix test' and 'mix compile' to validate changes within a worktree environment.
- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it reads external data from GitHub Action logs which could be manipulated by an attacker to include malicious instructions. 1. Ingestion points: Workflow run logs read via 'gh run view --log-failed' (SKILL.md). 2. Boundary markers: None; the skill does not instruct the agent to use delimiters or specific ignore-rules for content found in logs. 3. Capability inventory: The agent can modify code, manage git worktrees, and execute local project commands. 4. Sanitization: No specific sanitization of log content is defined, though the skill includes safety rules to stop and ask for human review if failures involve secrets, credentials, or architectural changes.
Audit Metadata