MCP Integration

Use ArchitectAI in Your IDE

Connect ArchitectAI to Antigravity, Cursor, Windsurf, or Claude Desktop via the Model Context Protocol. Generate UML diagrams and full SDLC projects from your IDE — all saved to your dashboard automatically.

View Setup Guide

One server file, dozens of tools — all in a single download. Run it on your machine and connect your IDE below.

No code required
2 min setup
Works with any IDE

Option 1 — Run on your machine

How it works

💻Your IDE

Antigravity / Cursor

⚙️MCP Server

runs on your machine

☁️ArchitectAI API

ai-uml-api.onrender.com

📊Your Dashboard

servicesasai.com

Setup in 4 Steps

No source code needed. Works for any ArchitectAI user.

1

Download the MCP Server

Download server.py and requirements.txt and place them in a folder called architectai-mcp.

📁 Folder structure you'll have:

architectai-mcp/
├── server.py          ← MCP server (download above)
├── requirements.txt   ← dependencies
└── .env               ← auto-created on first connect
2

Install Dependencies

Run this once in your terminal. Requires Python 3.9+.

Terminal
cd architectai-mcp
pip install -r requirements.txt

Only 4 packages are installed: mcp, httpx, pydantic, python-dotenv. No other dependencies.

3

Configure Your IDE

Select your IDE below and add the MCP config. Replace the path with the actual location of your server.py file.

1Open Antigravity Settings → search for "MCP Servers"
2Add the JSON config shown below to your MCP configuration
3Replace the path with the actual path on your machine
4Restart Antigravity — ArchitectAI tools will appear automatically
Antigravity — MCP Config
{
  "mcpServers": {
    "architectai": {
      "command": "python",
      "args": ["/absolute/path/to/architectai-mcp/server.py"]
    }
  }
}
⚠️

Important: Replace /absolute/path/to/architectai-mcp/server.py with the real path on your machine, e.g. /Users/alice/architectai-mcp/server.py on Mac or C:/Users/alice/architectai-mcp/server.py on Windows.

4

Connect Your Account

Open your IDE chat and type this once. Your browser will open — log in and your IDE is connected automatically.

IDE Chat Prompt
Connect my ArchitectAI account
🌐

Your browser opens → servicesasai.com/ide/connect

🔐

Sign in or create a free account

Your IDE is automatically authorized (30-day token)

🚀

You're ready! Start generating and saving diagrams

You're all set! The token expires after 30 days. Re-connect by saying "My ArchitectAI session expired. Please reconnect."

MCP Tools Reference

Dozens of tools in a single server — auth, diagrams, SDLC, prompts, and more

connect_to_saas

One-time browser auth flow. Opens the ArchitectAI login page, polls for your token, and saves it automatically. Run this first.

format_uml_prompt(code_snippet)

Formats a system prompt for your IDE's AI to generate Mermaid UML from a code snippet. The IDE runs the generation, ArchitectAI saves it.

save_diagram_to_saas(title, mermaid_code, type)

Saves a generated Mermaid diagram to your ArchitectAI dashboard. Returns a direct URL to view and share the diagram.

create_sdlc_project(name, description, requirement)

Creates a new SDLC project in your account with the raw business requirement. Returns a project_id for saving artifacts.

save_sdlc_artifacts(project_id, ...6 artifacts)

Saves all 6 SDLC phases: structured requirements, domain model, HLD, LLD, Mermaid diagrams, and validation report.

format_sdlc_prompt(stage, context)

Gets the exact system prompt for the IDE's AI to generate each SDLC phase. Stages: requirements, domain_model, hld, lld, diagrams, validation.

list_sdlc_projects()

Lists all your SDLC projects with IDs, names, statuses, and last updated dates.

get_sdlc_project(project_id)

Loads the full artifacts of an existing SDLC project. Use this to resume work or review results.

list_diagrams()

Lists all your saved UML diagrams with IDs, titles, and types.

get_diagram(diagram_id)

Loads a diagram by ID and returns its Mermaid code and metadata.

list_sdlc_templates() / list_sdlc_models() / list_sdlc_stacks()

Lists available SDLC templates, AI models, and code stacks.

format_requirements_prompt(context) … format_validation_prompt(context)

Six per-stage prompts: requirements, domain_model, hld, lld, diagrams, validation. Same content as format_sdlc_prompt(stage, context).

get_sdlc_project_artifacts_json(project_id)

Returns the full SDLC project and all artifacts as JSON.

ping_saas()

Checks connectivity to the ArchitectAI API (no auth required).

Example Prompts

Just type these in your IDE chat after connecting

🔵 Connect / Account

>

Connect my ArchitectAI account

📐 Generate a UML Diagram

>

Analyze src/auth.py and generate a sequence diagram showing the complete authentication flow. Save it to my ArchitectAI dashboard.

🗄️ ER Diagram from Models

>

Read all files in backend/models/ and generate a complete ER diagram showing every table, all columns, and foreign key relationships. Save it.

🚀 Full SDLC Project (one prompt)

>

Create a complete SDLC project for: "Build a task management app where users can create projects, assign tasks, set deadlines, and track progress." Generate all 6 phases and save everything to my ArchitectAI dashboard.

📋 List / Resume Projects

>

List all my SDLC projects on ArchitectAI.

🔄 Reconnect After Expiry

>

My ArchitectAI session expired. Please reconnect.

Troubleshooting

❌ Not connectedSay: "Connect my ArchitectAI account"
❌ Session expiredSay: "My ArchitectAI session expired. Please reconnect."
❌ Could not reach backendArchitectAI production API may be temporarily unavailable. Try again in 1-2 min.
❌ Free plan limit reachedUpgrade at servicesasai.com/billing for unlimited diagrams.
❌ Project not foundCheck your IDs by saying: "List all my SDLC projects"
Diagram link shows 404Log in at servicesasai.com first, then visit the link.
MCP tools not showing in IDEMake sure the path in your MCP config is absolute and server.py exists there. Restart the IDE.

Ready to connect your IDE?

Download the 2-file MCP server, install dependencies, add 5 lines of config, and you're live.