skills/bmad-code-org/bmad-method-test-architecture-enterprise/bmad-testarch-atdd/Gen Agent Trust Hub
bmad-testarch-atdd
Warn
Audited by Gen Agent Trust Hub on Apr 26, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill frequently invokes a Python script (
resolve_customization.py) located within the project's_bmad/scripts/directory to resolve workflow configurations and step logic. - [COMMAND_EXECUTION]: The completion steps in
steps-v/step-01-validate.mdandsteps-c/step-05-validate-and-complete.mdcontain anon_completehook that executes the return value of a customization script as a terminal command. This pattern allows for arbitrary command execution guided by the local configuration files. - [PROMPT_INJECTION]: In
steps-c/step-01-preflight-and-context.md, the workflow ingests untrusted data from user-provided story files ({story_file}) to drive the generation of API and E2E test scaffolds. The skill lacks explicit boundary markers or sanitization logic for this ingested content, creating an indirect prompt injection surface where adversarial story text could influence the generated test code or agent behavior.
Audit Metadata