Integrates with Ollama as a local LLM provider for natural language understanding and response generation.
Integrates with OpenAI's API as an LLM provider for natural language understanding and response generation.
Allows interaction with any Swagger/OpenAPI documented API, enabling natural language queries and automated workflow execution.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@swaggbot List all users with role admin"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Transform Swagger/OpenAPI docs into conversational interfaces. Self-hosted, local-first, AI-powered API exploration.
# Docker (recommended)
git clone https://github.com/techbloom-ai/swaggbot.git
cd swaggbot && cp .env.example .env
# Generate SESSION_SECRET and add LLM API keys
docker-compose up -dRelated MCP server: MCPHy
Swaggbot converts any Swagger/OpenAPI documented API into a conversational interface. Built for developers who want to:
Explore APIs naturally — Chat instead of reading raw JSON
Automate workflows — Multi-step operations with data extraction
Stay in control — Self-hosted, data stays local
Integrate anywhere — Web UI, MCP server, or direct API
🤖 Natural Language API Interaction — "List all users" → GET /users
🔄 Workflow Automation — Chain multiple API calls with dependency resolution
🔌 MCP Server — Use with Claude Desktop, Cursor, Windsurf
🏠 Self-Hosted — Your data, your infrastructure
🔐 Session Management — Encrypted auth tokens, per-session isolation
🌐 Multi-Provider LLM — Moonshot, OpenAI, Anthropic, Ollama
📊 Array Filtering — [name=John].id syntax for data extraction
🛡️ Security — CSP headers, rate limiting, encrypted storage
┌─────────┐ ┌─────────┐ ┌─────────┐
│ Web UI │ │ MCP │ │ API │
│ Next.js │ │ Server │ │ Clients │
└────┬────┘ └────┬────┘ └────┬────┘
└─────────────┼─────────────┘
▼
┌─────────────────┐
│ Chat Service │ Intent Classification
│ + LLM Provider │ → Curl Generation → Execution
└─────────────────┘
│
┌─────────────┼─────────────┐
▼ ▼ ▼
┌────────┐ ┌─────────┐ ┌──────────┐
│SQLite │ │Workflow │ │ Target │
│(Local) │ │ Engine │ │ API │
└────────┘ └─────────┘ └──────────┘Node.js 18+ or Docker
LLM API key (Moonshot, OpenAI, Anthropic, or Ollama)
# 1. Clone and configure
git clone https://github.com/techbloom-ai/swaggbot.git
cd swaggbot
cp .env.example .env
# 2. Set SESSION_SECRET (required for auth)
# Linux/macOS:
export SESSION_SECRET=$(openssl rand -base64 32)
# Or add to .env: SESSION_SECRET=your_random_secret_here
# 3. Add your LLM API key to .env
# MOONSHOT_API_KEY=your_key_here
# 4. Start
docker-compose up -dDatabase migrations run automatically on first startup.
git clone https://github.com/techbloom-ai/swaggbot.git
cd swaggbot
pnpm install
cp .env.example .env.local
# Edit .env.local with SESSION_SECRET and LLM API keys
pnpm db:migrate
pnpm devVariable | Required | Default | Description |
| Yes | — | Random secret for session encryption (min 32 chars) |
| Yes* | — | Moonshot AI API key |
| Yes* | — | OpenAI API key |
| Yes* | — | Anthropic API key |
| Yes* | — | Ollama server URL |
| No |
|
|
| No |
| SQLite database path |
| No |
| App base URL |
*At least one LLM provider required
# Linux/macOS
openssl rand -base64 32
# Or any random string (min 32 characters)All endpoints require authentication via session cookie.
Endpoint | Method | Description |
| POST | Authenticate and create session |
| POST | Create API session from Swagger URL |
| GET | List all sessions (paginated) |
| POST | Send message to API |
| POST | Create multi-step workflow |
| POST | Execute workflow |
{
"mcpServers": {
"swaggbot": {
"command": "docker",
"args": ["compose", "run", "--rm", "swaggbot-mcp"],
"env": {
"SESSION_SECRET": "your_secret",
"MOONSHOT_API_KEY": "your_key"
}
}
}
}Navigate to login page
Create a session with your Swagger URL
Start chatting:
"Create a user named John"
"List all pets with status available"
"Execute the login workflow"# Authenticate
curl -X POST http://localhost:3003/api/auth/login \
-H "Content-Type: application/json" \
-d '{"password": "your_password"}' \
-c cookies.txt
# Chat with API
curl -X POST http://localhost:3003/api/chat \
-H "Content-Type: application/json" \
-b cookies.txt \
-d '{"sessionId": "...", "message": "List all users"}'Component | Technology |
Framework | Next.js 16 (App Router) |
Language | TypeScript 5 |
Database | SQLite (libSQL/Turso) |
ORM | Drizzle ORM |
UI | shadcn/ui + TailwindCSS 4 |
State | Zustand |
LLM SDK | Vercel AI SDK patterns |
MCP | Model Context Protocol SDK |
Testing | Vitest |
Container | Docker + Docker Compose |
swaggbot/
├── app/ # Next.js App Router
│ ├── api/ # API routes
│ ├── sessions/ # Session UI
│ └── settings/ # Settings page
├── components/ # shadcn/ui components
├── lib/
│ ├── db/ # Database schema & migrations
│ ├── llm/ # LLM provider implementations
│ ├── services/ # Business logic
│ ├── auth/ # Session & encryption
│ └── prompts/ # LLM prompt management
├── scripts/ # MCP server & entrypoint
└── data/ # SQLite storage (Docker volume)Local-first: All data stored locally in SQLite
Encrypted tokens: Auth tokens encrypted with AES-256-GCM
Session-based auth: Password-protected access
CSP headers: Content Security Policy protection
Rate limiting: Per-endpoint limits configurable
No telemetry: Zero analytics or tracking
MIT © TechBloom
This server cannot be deployed
MCP server for AI access to Swagger by SmartBear.
MCP server unifying ERPs, CRMs, APIs and knowledge base for Claude, ChatGPT and Gemini.
MCP server that lets AI assistants use all OneSchema features exposed via the public API.
Hosted MCP server connecting AI assistants to 9,000+ apps and 40,000+ actions via Zapier.
