The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/sweep.py executes the macOS security utility via subprocess.run to retrieve the typesafe-api-key from the system keychain for API authentication. This is a documented administrative action for the skill.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data which is then used to construct prompts for an LLM, creating a vulnerability surface.
Ingestion points: Records are loaded from labelled.json and criteria configurations from configs.json in scripts/sweep.py.
Boundary markers: The script does not implement specific delimiters or 'ignore' instructions for the ingested data.
Capability inventory: Performs authenticated POST requests to https://api.typesafe.ai/v1/systemone using urllib.request.
Sanitization: The script does not sanitize the input strings before sending them to the remote API.

jev-eval