Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.pre.dev/llms.txt

Use this file to discover all available pages before exploring further.

The pre.dev MCP Server replaces Cursor Plan Mode with production-ready specification generation. Give your AI agents the architecture context they need to build complex systems correctly the first time.

​
Quick Start

1

Sign in to pre.dev

pre.dev account

Make sure you’re signed in. The MCP install will open a browser tab to ask which account (solo or enterprise organization) to authorize.
2

Install the MCP Server

Run this in your terminal:
claude mcp add --transport http predev https://api.pre.dev/mcp
A browser window opens — sign into pre.dev (if not already), pick which account or organization to authorize, and click Approve. Claude Code completes the handshake automatically.
For troubleshooting, see the Claude Code MCP documentation.
3

Test the Integration

Open your AI coding agent and try this command:
Use pre.dev to generate a spec for a task management app
If successful, you’ll receive a comprehensive project specification!

​
Why Your AI Agent Needs pre.dev

AI coding agents like Cursor are powerful, but they struggle with complex, multi-file projects without proper architecture. pre.dev solves this by generating comprehensive technical specifications that transform how your AI agent understands and builds systems.

​
The Problem with Current AI Planning

Standard AI planning modes (like Cursor Plan) provide basic task decomposition, but they:
  • ❌ Lack deep architectural context
  • ❌ Miss critical implementation details
  • ❌ Don’t consider database schemas, API contracts, or system boundaries
  • ❌ Force you to manually guide complex builds step-by-step

​
How pre.dev Makes Your AI Agent Better

The pre.dev MCP Server equips your AI coding agent with:
  • ✅ Enterprise-Grade Architecture: Complete system designs with database schemas, API contracts, and component hierarchies
  • ✅ Implementation Roadmaps: Detailed, sequenced build plans that guide your agent through complex projects
  • ✅ Technology Decisions: Pre-made architectural choices for auth, state management, APIs, and more
  • ✅ Full Context Understanding: Your agent knows exactly what to build, how to build it, and why
Result: Your AI agent goes from “helpful assistant” to “senior architect + developer” in seconds.

​
Available Tools

fast_spec

Cost: ~5-10 credits (variable) | Speed: ~30-60 secondsQuick project specifications perfect for rapid prototyping and MVP planning.
Use pre.dev fast_spec to generate a spec for:
A social media app with real-time messaging,
photo sharing, and user profiles

deep_spec

Cost: ~10-50 credits (variable) | Speed: ~2-5 minutesComprehensive, detailed specifications ideal for enterprise projects and complex systems.
Use pre.dev deep_spec to generate a detailed spec for:
An enterprise CRM with AI-powered lead scoring,
automated workflows, and advanced analytics

​
Advanced Features

Build on existing work by referencing previous specs or projects:
Use pre.dev deep_spec to add payment processing 
to this existing project: [existing_context_id]
The MCP will automatically:
  • Fetch your existing context
  • Validate permissions
  • Generate new specs that build on your previous work
Automatically scrape and archive external documentation alongside spec generation.Simply reference documentation URLs in your request:
Use pre.dev deep_spec to generate a spec for a payment system 
with Stripe integration. Reference https://stripe.com/docs/api
What you get:
  • ✅ Accurate Specs: AI uses the documentation to generate precise integration specifications
  • ✅ Documentation Archives: Downloadable ZIP files containing all scraped documentation
  • ✅ Organized by Platform: Each documentation source gets its own well-structured archive
  • ✅ Parallel Processing: Documentation scraping runs simultaneously with spec generation (no slowdown)
  • ✅ Graceful Fallback: Spec generation completes even if documentation scraping fails
Enterprise users can view and download documentation archives from the API Usage Logs in their dashboard.Example with multiple docs:
Use pre.dev deep_spec to generate a healthcare platform spec 
with HL7 FHIR integration and HIPAA compliance. 
Reference https://docs.hl7.org and https://www.hhs.gov/hipaa
You’ll receive separate archives for HL7 and HIPAA documentation alongside your spec.

​
Cost Management

Check Your Usage: Visit pre.dev/projects/playground to view remaining credits, usage history, and upgrade options.
Out of Credits?If you run out of credits, you’ll see this message:
❌ Insufficient Credits Available
You have insufficient credits to generate a spec.
Visit the playground to upgrade your plan or purchase additional credits.

​
Common Use Cases

Perfect for MVPs and quick validation:
Use pre.dev fast_spec to generate a spec for: 
MVP of a food delivery app with restaurant listings, 
order tracking, and payment integration

​
Troubleshooting

  • Re-run claude mcp add (or the Cursor one-click) to redo the browser authorization
  • Make sure you’re signed into the right pre.dev account in your browser before the consent page loads
  • If you’re an enterprise admin, pick the correct organization on the consent page — each org has its own credit pool
  • Check that your subscription is active at pre.dev/projects/playground
  • Restart your editor after adding the MCP config
  • Verify the URL: https://api.pre.dev/mcp
  • Check JSON syntax in configuration