The Agent Skills Directory

[SAFE]: The skill utilizes standard project inspection commands such as git status, git show, find, and grep to identify documentation needs and map project structure. These operations are benign and scoped to the local repository.
[SAFE]: Orchestration involves specialized agents (using Haiku, Sonnet, and Opus models) for analysis, technical writing, and quality review. This multi-step process includes a dedicated verification gate to ensure documentation accuracy and consistency.
[SAFE]: Indirect Prompt Injection Surface: The skill ingests untrusted data from the local codebase to generate documentation.
Ingestion points: Reads local code changes and existing documentation files through git commands and directory traversal.
Boundary markers: Not explicitly defined in the instruction templates; however, the use of specialized 'Quality Review' agents acts as a logical control.
Capability inventory: The skill is capable of reading file metadata, repository status, and writing documentation files to the local disk.
Sanitization: No explicit sanitization of code content is performed, relying instead on the specialized tech-writer and review agents to filter and process the information.
[SAFE]: No evidence of data exfiltration, credential harvesting, obfuscation, or unauthorized remote code execution was found. All operations align with the stated purpose of documentation maintenance.

update-docs