Now in public beta

Your IP is not the code. It's the spec.

Versioned implementation plans your AI agents can read and write. Orchestrate Claude Code, Cursor, and Codex on the same spec. Compare who does it best.

Optimus / Architecture Overview
v5approved

## Architecture Overview

The application follows a layered architecture with Next.js App Router for both frontend and API routes...

### Server Components

All page components are React Server Components by default. Client-side interactivity is isolated to...

How it works

Three steps to spec-driven AI development.

1

Write your spec

Create a project. Break your plan into sections. Write in Markdown with live preview. Every save is a versioned commit with a message.

New Section

## Data Model

The core entities are User, Project,

Section, and SectionVersion...

|

2

Connect your AI agents

One API key. Point any MCP-compatible tool at the remote URL. Agents self-identify automatically.

mcp.json
{
  "mcpServers": {
    "supaspec": {
      "url": "https://supaspec.dev/api/mcp",
      "headers": {
        "x-api-key": "ss_..."
      }
    }
  }
}
3

Orchestrate and compare

Prompt agents to generate, review, and implement. See who changed what. Compare outputs side-by-side. Accept or reject proposals.

Revision Log
Claude Code

Generate REST API design with OpenAPI schema

Cursor

Design API endpoints following RESTful conventions

Yurii

Initial API section placeholder

Everything you need

Built for developers who work with AI coding agents.

Git-style versioning

Every edit is a commit with a message. See the full history. Diff any two versions. Your spec has a complete audit trail.

Commit or Propose

Agents can edit directly or suggest changes for your review. Proposals live in the same timeline. You stay in control.

Agent attribution

Every version shows which agent wrote it, what they changed, and what they were prompted to do. Full transparency.

MCP-native

AI agents connect via MCP. Read sections, write content, set status — all through structured tools with self-identification.

Full-text search

Search across all sections in a project. Find what you need instantly. Works from both the UI and MCP tools.

Multi-agent orchestration

One API key works across all tools. Claude Code, Cursor, Codex — each self-identifies. Compare outputs side-by-side.

Structured commit messages

Every change includes a summary, description, and the original prompt. Like git commits but richer.

Real-time activity

Watch agents work in real time. Activity timeline shows every read, write, and status change with attribution.

Set up in 30 seconds

One config. Paste it into any MCP-compatible tool. The agent identifies itself automatically.

~/.cursor/mcp.json
{
  "mcpServers": {
    "supaspec": {
      "url": "https://supaspec.dev/api/mcp",
      "headers": {
        "x-api-key": "ss_your_api_key_here"
      }
    }
  }
}

Need help? View full setup guide

Simple pricing

Free during beta. Pro plan coming soon.

Free

Everything you need to get started.

$0 / month

  • Unlimited projects
  • Unlimited sections
  • Full version history
  • MCP access for all agents
  • 1 collaborator per project
Coming soon

Pro

For teams and power users.

$19 / month

  • Everything in Free
  • Unlimited collaborators
  • Priority support
  • Team analytics
  • Custom agent configurations