The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses bash to manage the QA environment lifecycle via stack up and stack down commands, and to execute the test suite.
[COMMAND_EXECUTION]: A local Python script workflows/validate_tasks.py is invoked to validate the integrity and schema of the generated tasks.json configuration file.
[EXTERNAL_DOWNLOADS]: The skill can delegate test generation to external skills (such as those provided by Anthropics) based on settings in the local skills.config.yaml file.
[PROMPT_INJECTION]: The skill processes untrusted user data including feature descriptions and acceptance_criteria to derive test cases.
Ingestion points: acceptance_criteria, feature, and contract_summary inputs (SKILL.md).
Boundary markers: None explicitly defined for input interpolation, though the skill instructions mandate following strict QA standards.
Capability inventory: Subprocess execution via stack, bash, and python3; file system writes for test specs and configuration (SKILL.md).
Sanitization: The skill includes a 'No weakened assertions' standard and configuration-level freezes (gates.freeze_approved_tests) to prevent adversarial manipulation of test outcomes.

qa-automation