Igniral MCP Server
Official# Igniral MCP Server
> **[Igniral](https://igniral.com)** ā Production-Ready Backends with AI Speed
Model Context Protocol (MCP) server that bridges AI agents (Claude, Cursor, Antigravity, etc.) with [Igniral's](https://igniral.com) backend platform. Describe your API in plain English ā Igniral generates the schema, CRUD endpoints, authentication, Swagger docs, and antivirus-protected file storage automatically.
## What is Igniral?
[Igniral](https://igniral.com) lets you generate 100% of your API infrastructure with a simple prompt, or build manually using a Visual Schema Builder. Everything is production-ready from the start:
- š¤ **AI-Powered Generation** ā Describe your data model, get a complete REST API instantly
- š **Built-in Auth & RBAC** ā JWT authentication with role-based access control, no auth code needed
- š **Always-Sync Swagger** ā OpenAPI docs update automatically with every change
- š”ļø **Antivirus File Storage** ā Every uploaded file is scanned by ClamAV before reaching your infrastructure
- š **Real-time Analytics** ā Monitor API usage, error rates, and traffic from your dashboard
- šļø **Managed Database** ā Automatic backups and replication, zero DBA required
**Get started for free** at [igniral.com](https://igniral.com) ā [Start Now](https://auth.igniral.com/subscribe)
## Prerequisites
1. **Create an Igniral account** at [igniral.com](https://auth.igniral.com/subscribe) (free tier available)
2. **Generate Agent API Keys** from the [Igniral Dashboard](https://auth.igniral.com/login) ā Agent API Keys
3. **Have Node.js ā„ 18** installed on your machine
> **Important:** This is an MCP server ā it runs inside your AI-powered IDE (Claude Desktop, Cursor, Antigravity, etc.), not directly from the terminal. You configure it once in your IDE settings, and the IDE handles starting and stopping it automatically.
## Quick Start
Choose your IDE and add the following configuration. Replace `agent-xxxxxxxxxxxx` and `your-client-secret` with your actual Agent API Key credentials.
### Claude Desktop
Edit your `claude_desktop_config.json`:
```json
{
"mcpServers": {
"igniral": {
"command": "npx",
"args": ["-y", "igniral-mcp-server"],
"env": {
"IGNIRAL_CLIENT_ID": "agent-xxxxxxxxxxxx",
"IGNIRAL_CLIENT_SECRET": "your-client-secret"
}
}
}
}
```
### Cursor
Edit your `.cursor/mcp.json`:
```json
{
"mcpServers": {
"igniral": {
"command": "npx",
"args": ["-y", "igniral-mcp-server"],
"env": {
"IGNIRAL_CLIENT_ID": "agent-xxxxxxxxxxxx",
"IGNIRAL_CLIENT_SECRET": "your-client-secret"
}
}
}
}
```
### Antigravity (Google)
Edit `~/.gemini/antigravity/mcp_config.json`:
```json
{
"mcpServers": {
"igniral": {
"command": "npx",
"args": ["-y", "igniral-mcp-server"],
"env": {
"IGNIRAL_CLIENT_ID": "agent-xxxxxxxxxxxx",
"IGNIRAL_CLIENT_SECRET": "your-client-secret"
}
}
}
}
```
> **Note:** Antigravity may not inherit your shell's `PATH`. Use the absolute path to `node` (e.g., `/opt/homebrew/Cellar/node/25.9.0_2/bin/node`) if you get "executable not found" errors.
That's it! After saving the configuration and restarting your IDE, you can ask your AI agent things like:
- *"Build me a gym management API"*
- *"Create a REST API for a pet store with products, orders, and users"*
- *"List my existing Igniral applications"*
## Tools
Once configured, the following tools are available to your AI agent:
| Tool | Description |
|------|-------------|
| `igniral_generate_schema_from_prompt` | Auto-generate a complete app from a natural language description |
| `igniral_create_application` | Create an empty application shell manually |
| `igniral_create_dynamic_endpoint` | Add API endpoints to an existing application |
| `igniral_list_applications` | List the user's existing applications |
## Architecture
```
AI Agent (Claude/Cursor)
ā
ā¼ (MCP Protocol - stdio)
āāāāāāāāāāāāāāāāāāāāāāāāāāāā
ā Igniral MCP Server ā
ā āā TokenManager ā ā OAuth2 client_credentials
ā āā Zod Validation ā
ā āā SSE Client ā
ā āā HTTP Client ā
āāāāāāāāāāāāāā¬āāāāāāāāāāāāāā
ā (HTTP + JWT with sub=userId)
āāāāāāāāā¼āāāāāāāā
ā¼ ā¼ ā¼
auth- ai-schema- json-elements
server builder microservice
(token) (auto-gen) (CRUD)
```
## Environment Variables
| Variable | Required | Description |
|----------|----------|-------------|
| `IGNIRAL_CLIENT_ID` | ā
| Agent API Key client ID (from Dashboard) |
| `IGNIRAL_CLIENT_SECRET` | ā
| Agent API Key client secret (shown once at creation) |
| `IGNIRAL_AUTH_URL` | ā | Auth server URL (default: `https://auth.igniral.com`) |
| `IGNIRAL_API_URL` | ā | API URL (default: `https://api.igniral.io`) |
| `IGNIRAL_AI_API_URL` | ā | AI API URL (default: `https://ai.igniral.com`) |
## Development (from source)
Only needed if you want to contribute or modify the server:
```bash
git clone https://github.com/igniral/igniral-mcp-server.git
cd igniral-mcp-server
npm install
cp .env.example .env # Edit with your credentials
npm run dev # Start in development mode
npm run inspect # Test with MCP Inspector
```
## Links
- š **Website:** [igniral.com](https://igniral.com)
- š **Documentation:** [igniral.com/docs](https://igniral.com/docs)
- š¬ **Community:** [Discord](https://discord.gg/ZrMbjPJh8w)
- š§ **Support:** [support@igniral.com](mailto:support@igniral.com)
## License
MIT
TDQS
Scored across 5 tools
Each tool targets a distinct action: creating an empty app, adding endpoints, generating full app from prompt, listing apps, and updating endpoints. The descriptions explicitly warn against using create_application after generate_schema_from_prompt, reducing ambiguity.
All tools follow a consistent verb_noun pattern with the prefix 'igniral_': create_application, create_dynamic_endpoint, generate_schema_from_prompt, list_applications, update_dynamic_endpoint. Minor length variation does not break consistency.
5 tools is well-scoped for the domain of application and dynamic endpoint management. It covers creation (two modes), listing, and updating without being overly numerous or sparse.
The set covers creation, listing, and update operations but lacks any delete functionality (e.g., delete endpoint or delete application). This is a notable gap that could hinder agent workflows requiring removal.