harness-testing
Pass
Audited by Gen Agent Trust Hub on Sep 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides legitimate software engineering guidance for harness testing, emphasizing non-invasive techniques and interception seams.
- [SAFE]: Mentions the external repository celesteanders/harness as background reading; no automated downloads or execution occur.
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the processing of user-provided code to generate test harnesses, which presents an attack surface for indirect prompt injection. 1. Ingestion points: User-provided source code for handlers and system units as described in SKILL.md. 2. Boundary markers: No explicit boundary markers or instructions to ignore embedded prompts in the user's code are defined. 3. Capability inventory: The agent implementation of tests involves file system operations (write_to_file) and dependency mocking. 4. Sanitization: The skill does not provide specific validation or sanitization logic for the input source code.
Audit Metadata