git-workflow
Pass
Audited by Gen Agent Trust Hub on Feb 19, 2026
Risk Level: SAFE
Full Analysis
- [SAFE] (SAFE): The skill documentation and reference guides promote legitimate and secure version control practices.
- [DATA_EXFILTRATION] (SAFE): The skill incorporates clear rules against committing sensitive data like .env files and credentials, providing a strong baseline for secret prevention.
- [COMMAND_EXECUTION] (SAFE): All referenced commands are standard Git and Graphite CLI operations intended for local development and repository management.
- [PROMPT_INJECTION] (SAFE): No evidence of adversarial instructions or attempts to bypass agent safety filters was found in the text.
- [Indirect Prompt Injection] (SAFE): The skill possesses a potential ingestion surface where the agent analyzes code diffs during the 'split' workflow. However, the presence of safety guardrails and the requirement for manual verification mitigate this surface. 1. Ingestion points: 'scripts/git-workflow split' analyzes diffs. 2. Boundary markers: None explicitly defined. 3. Capability inventory: Local command execution for Git and Graphite. 4. Sanitization: The skill mandates manual verification of staged changes ('Always verify staged files before committing').
Audit Metadata