qa
Pass
Audited by Gen Agent Trust Hub on Aug 16, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXPOSURE_AND_EXFILTRATION]: The skill accesses codebase files to provide context for bug reports. However, it includes explicit instructions to avoid leaking internal implementation details, such as file paths or line numbers, in the final GitHub issues. Information is sent only to the GitHub CLI tool (
gh issue create) for the intended purpose of reporting bugs. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input describing software issues.
- Ingestion points: User input during the conversational QA session defined in
SKILL.md. - Boundary markers: The agent uses structured Markdown templates (e.g.,
## What happened,## Steps to reproduce) to delimit content. - Capability inventory: Uses GitHub CLI (
gh issue create) and codebase exploration (via Explore subagent). - Sanitization: The agent is explicitly instructed to rephrase user input into project-specific domain language and focus on behavioral descriptions rather than raw input, which helps prevent the propagation of malicious payloads.
Audit Metadata