qa
Pass
Audited by Gen Agent Trust Hub on Jun 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the GitHub CLI (
gh issue create) to programmatically create issues on a remote repository based on information gathered during the session. - [PROMPT_INJECTION]: An indirect prompt injection surface is present because the skill processes untrusted user input and codebase content (via the
Exploresubagent) to generate the content of GitHub issues. The risk is managed by instructional constraints that forbid the inclusion of file paths, line numbers, or internal implementation details, favoring domain language instead. - Ingestion points: User descriptions of problems and codebase context gathered by a subagent.
- Boundary markers: Absent; the instructions do not specify the use of delimiters when processing external descriptions.
- Capability inventory:
gh issue create(Write/Network operation). - Sanitization: The prompt provides explicit negative constraints against citing internal technical details, which serves as a form of output sanitization to prevent accidental data exposure.
Audit Metadata