Black Friday: Get 40% off Craft for Life!

用 Craft 創想

Living Documentation Hub

This workflow was submitted during our first hackathon by one of our users
Living Documentation HubLiving Documentation Hub

Turn your technical documentation into an AI-queryable knowledge base – developers ask questions in their IDE and get instant answers across your entire microservice architecture.

自己動手構建

As the creator says:
“It turns documentation into a living system that devs can query directly with AI for instant answers across microservices. Saved ~10× dev time in our tests.”

步驟1 – Organize Your Documentation in Craft

Create a hierarchical documentation structure in Craft:

  • Main overview page – High-level architecture and system design
  • Subpages per service – One page for each microservice with:
    • Service description and purpose
    • API endpoints and schemas
    • Database models
    • Key implementation details
    • Dependencies and connections
    • Example code and usage patterns

Structure it consistently so AI can parse it reliably.

步驟2 – Include Comprehensive Details

For each service, document:

  • What it does – Clear description of the service’s purpose
  • How it connects – Relationships to other services
  • API documentation – Endpoints, parameters, responses
  • Database schema – Tables, relationships, key fields
  • Code examples – Real examples of how to use it
  • Common patterns – How developers typically interact with it

步驟3 – Enable Craft MCP

Set up Craft’s MCP (Model Context Protocol) connection to Claude. This allows Claude to read your entire documentation structure through your IDE.

步驟4 – Set Up Cursor with Claude

In Cursor (or your preferred IDE with Claude integration):

  • Enable Claude as your AI assistant
  • Connect it to your Craft MCP documentation
  • Configure it to read your documentation when answering questions

步驟5 – Create Custom Instructions for Claude

Add instructions to your Cursor setup:

You have access to our complete technical documentation in Craft via MCP. When a developer asks a question:

1. Search the documentation for relevant services and code
2. Provide the complete context (which services are involved, how they connect)
3. Show exact code examples from the documentation
4. Explain the flow across services
5. Suggest the affected files and changes needed

步驟6 – Start Asking Questions

Developers can now ask questions directly in their IDE:

→ Claude reads Craft docs → Shows complete flow from frontend through backend to PDF generation

→ Claude reads Craft docs → Lists all affected services and provides ready code

→ Claude reads Craft docs → Points to exact service, file, and code

點擊範例即可將提示複製到剪貼簿

步驟7 – Keep Documentation Updated

Maintain your Craft documentation as your codebase evolves:

  • Update service descriptions when architecture changes
  • Add new API endpoints as they’re created
  • Document new patterns and common solutions
  • Keep examples current with your actual code

The documentation stays the source of truth, and Claude always has the latest context.

Impact the Creator Notices

TaskBeforeAfterSpeed Gain
Info lookup45 min3 min15x faster
Feature implementation2 hours10 min12x faster
Debugging3 hour20 min9x faster
Onboarding new devs2-4 weeks2-3 days7x faster

The Result

Your Craft documentation becomes a team intelligence hub – an AI-queryable knowledge base that developers can ask questions to directly from their IDE. Instead of searching, asking colleagues, or reading through code, developers get instant, accurate answers with full context across your entire microservice architecture.

As the creator says:
“Turns Craft into a team intelligence hub — an AI-queryable knowledge base for microservices. 10–15× productivity boost, 3–4 hours saved daily per developer, faster onboarding & better code quality.”

更多想法供您嘗試

返回畫廊