heuristic-testing
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's workflow requires the agent to ingest and process external, potentially untrusted project data including requirement documents, specifications, and source code.
- Ingestion points: The agent is instructed to "actively read relevant Spec, code, and existing tests" and "collect business, system, effective Spec and existing test materials" as documented in SKILL.md and references/artifact-routing.md.
- Boundary markers: The instructions do not define explicit delimiters or warnings to ignore embedded instructions within the processed project files, which could lead to the agent following malicious commands hidden in requirements or comments.
- Capability inventory: The skill possesses significant capabilities, including the ability to generate code ("writing-unit-tests"), execute test sessions ("running-test-sessions"), and write to the file system to maintain test artifacts ("testing/TCO.md").
- Sanitization: There are no specified procedures for sanitizing or validating the content of external project files before they are integrated into the agent's decision-making process.
Audit Metadata