The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted project configuration data which could be crafted to influence agent behavior. \n
Ingestion points: scripts/validate.py reads content from package.json, biome.json, and CSS files. \n
Boundary markers: Data is parsed using json.loads and regular expressions; no specific security delimiters are used when reporting findings back to the agent. \n
Capability inventory: The skill uses subprocess.run in scripts/validate.py to check the Bun version and in tests/test_validate.py for testing. \n
Sanitization: Basic validation is performed via JSON structure checks and regex pattern matching.\n- [COMMAND_EXECUTION]: The script executes shell commands to verify the environment. \n
Evidence: scripts/validate.py calls subprocess.run(['bun', '--version']) to verify Bun is installed.\n- [EXTERNAL_DOWNLOADS]: The skill provides instructions for downloading and executing a setup script from the well-known Bun service. \n
Evidence: scripts/validate.py contains a suggested fix: curl -fsSL https://bun.sh/install | bash.

stack