The Agent Skills Directory

[SAFE] (SAFE): The skill consists entirely of markdown documentation, templates, and configuration examples. No malicious patterns, obfuscation, or unauthorized data access were detected.
[COMMAND_EXECUTION] (INFO): The skill describes how to configure tool permissions for bash commands (e.g., git status) as examples for other agents but does not execute these commands itself.
[DATA_EXFILTRATION] (INFO): Lists local configuration file paths (e.g., ~/.config/opencode/agent/) for organizational purposes to guide the user on where to store agent definitions; no exfiltration behavior is present.

writing-agents