acceptance-first
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The validation script
scripts/check-fixture.pyutilizessubprocess.runto execute Python scripts (runner.py) in a local temporary environment to verify the workflow functionality. - [DYNAMIC_EXECUTION]: The script
scripts/check-fixture.pyprogrammatically generates multiple Python files (api.py,store.py,runner.py) from predefined string templates during the testing process. - [INDIRECT_PROMPT_INJECTION]: The skill instructions direct the agent to ingest and execute content from external files like "authoritative agreements" and "acceptance examples."
- Ingestion points:
SKILL.md(Step 1) requires reading external project documentation and agreement files. - Boundary markers: There are no specific instructions or delimiters used to prevent the agent from following malicious commands that might be embedded in the examples or criteria.
- Capability inventory:
SKILL.md(Step 3) explicitly instructs the agent to "Execute the acceptance examples," granting the agent the capability to run potentially untrusted code found in the project context. - Sanitization: No validation or sanitization of the external examples is performed before execution.
Audit Metadata