The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes shell commands to handle background tasks and process lifecycle management. It launches an HTTP server using python3 -m http.server 8000 & and provides commands to terminate processes via kill $SERVER_PID or lsof -ti:8000 | xargs kill -9.
[DATA_EXFILTRATION]: By starting an HTTP server with python3 -m http.server without an explicit bind address, the server listens on all network interfaces (0.0.0.0). This configuration can expose the contents of the repository to other devices on the same network.
[PROMPT_INJECTION]: Indirect Prompt Injection (Category 8) surface identified in the catalog parsing logic.
Ingestion points: The skill ingests untrusted data from index.json to identify file paths and URLs for verification.
Boundary markers: None; the agent does not use delimiters or instructions to ignore potentially malicious content within the JSON file.
Capability inventory: Includes filesystem existence checks using pathlib.Path and network connectivity tests via HTTP HEAD requests.
Sanitization: The provided Python snippet uses string splitting and path construction (Path(path).exists()) without validation, creating a surface for directory traversal if the index.json file contains malicious path sequences like ../../.

verify-catalog