The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions direct the agent to execute a bundled script, scripts/repo_test_probe.py, to analyze the project structure and testing environment. The script is a diagnostic tool that reads project metadata (e.g., package.json, pyproject.toml) and filesystem structure to recommend test commands. This behavior is transparent and consistent with the skill's primary purpose of test generation.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection because the repo_test_probe.py script reads and parses external configuration files from the repository being analyzed.
Ingestion points: The script reads content from package.json, pyproject.toml, and scans filenames throughout the repository.
Boundary markers: None are explicitly used when the agent reads the script's JSON output.
Capability inventory: The agent is encouraged to execute shell commands (the diagnostic script itself and discovered test runners like npm run test or pytest).
Sanitization: The script uses standard JSON parsing and string matching, but does not sanitize the contents of the configuration files before including them in the output report.

write-tests