The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides code snippets demonstrating how to perform network operations using the fetch API to interact with PinMe's internal endpoints for email and LLM services.
[CREDENTIALS_UNSAFE]: Guidelines describe using an environment variable API_KEY for service authentication. No hardcoded credentials or secrets are present in the documentation.
[DATA_EXFILTRATION]: Communication is directed to pinme.cloud, which is the official platform domain associated with the skill's functionality.
[PROMPT_INJECTION]: The example code for LLM integration ingests user-supplied text from request bodies, creating a surface for indirect prompt injection where malicious input could influence model behavior. \n
Ingestion points: handleChat and handleChatStream functions in SKILL.md read user input from the request body. \n
Boundary markers: Not utilized in example code; user input is directly mapped to message content. \n
Capability inventory: The code examples demonstrate network access to LLM APIs via fetch. \n
Sanitization: No sanitization or filtering logic is included in the implementation examples.

pinme-api