research

Documenting an existing codebase architecture involves using parallel sub-agents to explore components and interactions, explaining current implementation and file locations without suggesting changes, and synthesizing findings into structured Markdown reports with GitHub permalinks.

Generating technical documentation for an unfamiliar codebase is best handled by deploying specialized agents to comprehensively explore components and synthesize findings into structured Markdown, providing an objective overview of the module's implementation and system interactions.

Parallel sub-agent research for codebase documentation works by deploying specialized agents like 'scout' and 'thoughts-locator' to comprehensively explore current implementations and historical context, synthesizing structured Markdown reports with GitHub permalinks.

Yes, you can leverage a `thoughts/` directory to add historical context to codebase documentation. The research process uses this directory to incorporate past explorations and historical background alongside current implementation details.

Yes, codebase documentation includes GitHub permalinks for file locations. The research process synthesizes findings into structured Markdown reports that reference exact implementation file locations using GitHub permalinks.

You should avoid automated codebase documentation when you need code refactoring suggestions or architectural improvements, as this approach focuses strictly on explaining the current implementation objectively without recommending changes.