The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill follows security best practices for secret management by reading the required API_KEY from the environment. It explicitly instructs the agent never to print or log the key and avoids sourcing shell startup files, reducing the risk of accidental exposure.
[COMMAND_EXECUTION]: Uses standard system utilities including curl for network communication, jq for JSON processing, and uuidgen for idempotency. These commands are used as intended for the skill's research functionality.
[DATA_EXFILTRATION]: Sends user-provided research queries to an external service (deepapi.co). This is the primary function of the skill. It uses jq --rawfile to safely incorporate the research prompt into the JSON request body, which prevents malformed data or simple command injection within the JSON structure.

deep-research