Building Docs for the AI Era, Part 2: Inki, a Multi-Agent Documentation Pipeline

Strapi’s documentation team uses an open-source Claude Code plugin called Inki to turn feature requests or code pull requests into researched, drafted, reviewed, and submitted documentation updates, while preserving human approval at key stages. Designed around the actual documentation workflow rather than a single “write docs” command, the system separates research, writing, review, and submission into composable skills, with specialized agents handling tasks such as routing content to the correct existing page, generating outlines, drafting prose, validating code examples, checking cross-page consistency, and identifying known pitfalls. Its handoffs are readable documents containing both human-oriented explanations and machine-readable data, allowing writers to intervene or continue work at any point. The system emphasizes explicit uncertainty, read-only review agents, cost-appropriate model selection, and automated fixes only for unambiguous issues. Its central value lies less in the AI agents themselves than in the documented templates, area-specific authoring guides, Git conventions, and catalog of past mistakes that encode organizational knowledge in reusable Markdown files. Although Inki reduces time spent on placement, boilerplate, and verification, human writers remain responsible for prioritization, product context, upstream feedback, and pedagogical decisions about what readers need to learn and in what order.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.