The Agent Skills Directory

[COMMAND_EXECUTION]: The bundled script scripts/inspect_test_health.py uses the subprocess.run function to execute git commands (e.g., git log, git rev-parse).
Evidence: The script calls subprocess.run(["git", "-C", str(root), ...]) to extract repository history and state. These calls do not use shell=True and are limited to the git binary, which is a standard requirement for repository analysis skills.
[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and analyze untrusted data from the repository being audited, creating a surface for indirect prompt injection.
Ingestion points: The script inspect_test_health.py reads the content of configuration files (e.g., package.json, pyproject.toml, CI YAML files) and samples test files.
Boundary markers: The instructions do not provide specific boundary markers or warnings to the agent to ignore instructions contained within the audited files.
Capability inventory: The agent is encouraged to execute shell commands found in the repository's manifests (e.g., npm test, pytest) as part of the assessment workflow.
Sanitization: There is no evidence of sanitization or filtering of the repository content before it is processed by the agent.

test-health