The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local Python scripts (resolve_customization.py, resolve_config.py) and standard tools (uv, npx, playwright-cli) located within the project's own _bmad directory or official package registries. These are used for environment configuration and evidence collection as part of the primary test review purpose.
[EXTERNAL_DOWNLOADS]: The skill may prompt the user to install a missing knowledge base from the author's official GitHub repository (bmad-code-org/bmad-method-test-architecture-enterprise). This is a transparent installation of a required dependency from a verified source.
[DYNAMIC_EXECUTION]: The skill uses a 'subagent' architecture where the orchestrator dispatches specific quality checks (Determinism, Isolation, etc.) to worker files. This is a modular design pattern for parallel processing of analysis tasks and does not involve the execution of untrusted or remote code.
[INDIRECT_PROMPT_INJECTION]: The skill analyzes external test files and context artifacts (stories, PRDs). It implements strong boundary protections by explicitly stating that context files are read for understanding only, are never reviewed or scored, and can never waive a violation or lower a severity. This mitigates risks from malicious instructions embedded in the processed data.

bmad-testarch-test-review