qe-test-execution
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted repository metadata, such as git changes and pull request descriptions, to dynamically construct and execute test suites. This functionality creates an inherent attack surface where malicious file names or metadata in a repository could attempt to influence the agent's behavior during test selection and execution orchestration.
- Ingestion points: Data enters the agent's context through pull request analysis (PR #123) and git change detection (gitChanges) as shown in the workflow examples in SKILL.md and the test cases in evals/qe-test-execution.yaml.
- Boundary markers: There are no explicit delimiters or specific instructions for the agent to ignore or sanitize embedded instructions within the ingested code metadata.
- Capability inventory: The skill uses the 'aqe' tool to execute shell commands with arguments derived from the analyzed repository data.
- Sanitization: The skill instructions do not describe a validation or sanitization process for the file paths or test identifiers extracted from the repository before they are used in command-line arguments.
Audit Metadata