The Agent Skills Directory

[COMMAND_EXECUTION]: The skill defines a PreToolUse hook in SKILL.md that executes scripts/guard-test-integrity.sh whenever the Edit or Write tools are invoked. The hook searches for the script in standardized locations including the plugin root and project directory.
[COMMAND_EXECUTION]: The guard-test-integrity.sh script executes logic via python3 -c to parse tool call JSON and identify specific test-runner keywords. This is used to enforce the integrity of the test suite by blocking the addition of suppression markers like it.skip or pytest.mark.skip.
[PROMPT_INJECTION]: The skill processes untrusted project code and test files to identify missing behaviors and generate test cases. \n
Ingestion points: Project source files read through Read, Glob, and Grep tools, and file content updates captured by the PreToolUse hook. \n
Boundary markers: The skill instructions provide clear directives to maintain test integrity, and the automated hook provides a deterministic enforcement boundary. \n
Capability inventory: The skill is granted access to local shell execution and file modification tools. \n
Sanitization: The integrity hook uses regular expressions to identify and filter out instructions that would weaken the test signal, such as framework-specific skip directives or linting suppressions.

auto-test