close-new-library-support-pr
Pass
Audited by Gen Agent Trust Hub on Jun 24, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the GitHub CLI (
gh) and its GraphQL/REST APIs to perform repository operations, including querying pull request metadata, modifying issue assignees, and updating project board statuses. - [COMMAND_EXECUTION]: Includes specific instructions for permission escalation via
gh auth refresh -s projectif the current environment lacks the necessary scopes to modify GitHub Project boards. This is consistent with the skill's stated purpose of managing project items. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted content from external pull requests and issues.
- Ingestion points: Pull request bodies, titles, and issue descriptions are ingested using
gh pr viewand GraphQL queries in files likeSKILL.md. - Boundary markers: The instructions do not specify the use of delimiters or 'ignore' commands when reading data from GitHub pull requests or issues.
- Capability inventory: The skill possesses capabilities to modify repository state, including closing pull requests (
gh pr close), removing assignees (gh issue edit), and changing project statuses (gh project item-edit). - Sanitization: There are no instructions for sanitizing or escaping the retrieved GitHub content before it is processed by the agent to make state-changing decisions.
Audit Metadata