The Agent Skills Directory

[SAFE]: The skill implements a robust five-stage workflow that requires explicit user confirmation at critical steps, including scope definition, content outline review, and final file writing path.
[SAFE]: The skill operates entirely on the local codebase and does not initiate network requests, download external scripts, or interact with remote servers.
[SAFE]: Writing guidelines specifically instruct the agent to 'Describe, never copy' source code, which mitigates the risk of leaking sensitive logic into documentation and serves as a natural filter against processing malicious code as instructions.
[SAFE]: The instructions do not contain any prompt injection attempts, obfuscated content, or persistence mechanisms. Metadata and license information are standard and transparent.

describe-design