enterprise-testing
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute various local shell commands for testing, including
node --test,pnpm test, and a repository-specific scriptscripts/ci/run-tests.mjs. - [INDIRECT_PROMPT_INJECTION]: The diagnostic functionality requires the agent to fetch and interpret CI logs using the
ghCLI. These logs are generated by build processes that may process code from untrusted contributors, creating a surface where malicious instructions could be embedded in the output to influence the agent. - Ingestion points: CI logs fetched via
gh run view <run-id> --job <job-id> --logand PR metadata. - Boundary markers: None specified for the interpretation of external log content.
- Capability inventory: The agent has the ability to execute repository scripts and shell commands (
node,pnpm,gh). - Sanitization: No sanitization or filtering of log content is mandated before analysis.
Audit Metadata