Docs

Fan out documentation work across independent output files, per workflows/parallel-execution.md.

--init = first-time generation. --update (default) = refresh against the current codebase. No flag and ./docs/ missing or empty → --init.

Additional requests

<additional_requests> $ARGUMENTS </additional_requests>

Wave 0 (serial, 1 docs-manager run)

Generate the repomix-output.xml compaction + read existing ./docs and README.md. This is the shared baseline every wave-1 instance reads — do NOT re-run repomix per instance.

Wave 1 (parallel, ≤4 per workflows/parallel-execution.md §4)

Spawn one docs-manager instance per output file, each scoped to write ONLY its file from the wave-0 baseline and given <additional_requests> for context.

--init:

• Batch 1 (4): docs/project-overview-pdr.md, docs/codebase-summary.md, docs/code-standards.md, docs/system-architecture.md
• Batch 2 (1): README.md (keep ≤300 lines)

--update:

• Batch 1 (4): README.md, docs/project-overview-pdr.md, docs/codebase-summary.md, docs/code-standards.md
• Batch 2 (≤4): docs/system-architecture.md, docs/project-roadmap.md, docs/deployment-guide.md [optional, skip if absent], docs/design-guidelines.md [optional, skip if absent]

Print the ## Parallel Execution Matrix table (Unit/Owns/DependsOn/Wave/Verify — schema in workflows/parallel-execution.md §1) before delegating. Barrier: all wave-1 instances report before finalizing (§5).

Important

• Use docs/ as the source of truth for documentation.
• Read-only summary instead of a rewrite? Use /vk-docs-summarize.