gh
Pass
Audited by Gen Agent Trust Hub on Feb 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill documents numerous shell commands using the GitHub CLI (
gh). These are used for administrative and development tasks on GitHub. - [CREDENTIALS_UNSAFE]: Includes the
gh auth tokencommand, which displays the active authentication token. This poses a potential risk of credential leakage if the agent's output is logged or shared. - [PROMPT_INJECTION]: Reading content from pull requests and issues creates an indirect prompt injection surface where malicious external data might influence agent behavior.
- Ingestion points: Data enters via commands like
gh pr list,gh issue list, andgh api. - Boundary markers: The skill does not define specific delimiters to wrap or isolate content retrieved from GitHub.
- Capability inventory: The skill allows for repository creation, merging PRs, and running workflows via the
ghtool. - Sanitization: No explicit sanitization of data retrieved from external GitHub sources is documented.
Audit Metadata