The Agent Skills Directory

[COMMAND_EXECUTION]: The skill operates by executing the kuroco-admin CLI tool through shell commands to interact with the Kuroco management API.
Evidence: The skill documentation (SKILL.md) provides detailed patterns for using Bash tool calls to run commands such as kuroco-admin exec, kuroco-admin help, and kuroco-admin login.
[SAFE]: The skill incorporates security best practices for handling sensitive session information and credentials.
Evidence: The instructions explicitly forbid the agent from displaying or logging cookie values stored in ~/.kuroco-admin/cookies.txt and caution against sharing --verbose output which contains HTTP headers.
[SAFE]: Implementation of human-in-the-loop protocols for sensitive operations.
Evidence: The skill enforces a rule requiring the agent to obtain explicit user confirmation before performing write or delete operations (INSERT, UPDATE, DELETE).
[SAFE]: The skill identifies an attack surface for Indirect Prompt Injection but mitigates it through structured output and verification.
Evidence: The advise command processes natural language but returns structured JSON with system-verified endpoints and API specifications to prevent hallucination or malicious instruction following.

kuroco-admin-api