AI vibe coding has made individual developers incredibly productive. Claude Code, Cursor, Copilot — there are already powerful tools for writing code on your own.

But real software is never built alone.

Working in a team means aligning on "what to build" before writing any code. You gather requirements, discuss specs, define acceptance criteria, get reviews, and only then start implementing. Most AI development tools don't address this upstream collaboration — the discussion, alignment, and spec writing that happens before code.

That's where Colign comes in.

• Discuss and write specs together with AI
• Co-edit specs in real-time with your team
• Structure the alignment process with a Draft → Design → Review → Ready workflow
• Once specs are finalized, AI helps implement based on them

AI that writes code already exists. Colign makes sure your team is looking at the same thing before the code gets written.

• AI Spec Generation — Generate structured specs from a single prompt using your own API key
• Real-time Co-editing — Collaborate on specs simultaneously with your team
• Structured Proposals — Problem, Scope, Approach, Acceptance Criteria in a consistent format
• Project Memory — Persistent context (domain rules, constraints, decisions) shared across all specs
• Workflow States — Draft → Design → Review → Ready pipeline for every change
• MCP Server — Connect Claude Code, Cursor, or any MCP-compatible AI tool to read/write specs
• Dashboard & Inbox — Track spec status, reviews, and notifications in one place

Colign follows a Spec-Driven Development (SDD) approach. When AI can generate code in minutes, the bottleneck shifts from "can we build it?" to "have we defined it correctly?"

Traditional PRDs tried to put everything in one 30-page document. Colign splits this into two layers:

Project Memory — Strategic context that rarely changes. Domain rules, business constraints, target users, and technical decisions. Written once, referenced by every Change.

Structured Proposal — Tactical spec for each Change. Lightweight, structured, and designed for both humans and AI agents to read:

1. Platform AI — Use your own API key to generate structured specs directly in Colign
2. External AI — Connect Claude Code, Cursor, or other AI tools via MCP Server to read/write specs in Colign

For design decisions and competitive analysis behind this structure, see docs/structured-proposal.md.

Colign exposes an MCP (Model Context Protocol) server so any AI tool — Claude Code, Cursor, Windsurf, VS Code Copilot — can read and write specs directly.

No binary to install. Just add the URL and your API token:

{
  "mcpServers": {
    "colign": {
      "url": "https://api.colign.co/mcp",
      "headers": {
        "Authorization": "Bearer col_your_token_here"
      }
    }
  }
}

Generate an API token at Settings > AI & API Keys in the Colign web app.

For local development or self-hosted instances:

go build -o colign-mcp ./cmd/mcp

COLIGN_API_TOKEN=col_... COLIGN_API_URL=http://localhost:8080 ./colign-mcp

Colign ships with a Claude Code plugin that adds workflow skills on top of MCP.

# 1. Register marketplace
/plugin marketplace add https://github.com/colign/plugin

# 2. Install plugin
/plugin install colign@colign

# 3. Reload and connect (browser opens for OAuth login)
/reload-plugins

6 skills that follow the change lifecycle:

onboard → explore → propose → plan → implement → complete

Skills trigger automatically by context (e.g., "implement the next task") or explicitly via /colign:implement.

# Start all services (API + DB + Redis)
docker-compose up --build

# Run frontend (separate terminal)
cd web && npm install && npm run dev

Open http://localhost:3000 to access the app.

# Generate proto (Go + TypeScript)
cd proto && buf generate

# Run API server locally
go run ./cmd/api

# Run tests
go test ./...

• Go 1.26+
• Node.js 20+
• Docker & Docker Compose
• buf

This project is licensed under the GNU Affero General Public License v3.0 (AGPL-3.0).