Metorial
ActiveDescription
Metorial is a unified integration layer that connects any AI model to 1200+ MCP servers, CLIs, and APIs through a single SDK, eliminating the need to wire each tool into your agent.
Key Features
- 1200+ integrations — One-click enable MCP servers, CLI tools, and REST APIs without per-tool wiring
- Single SDK — A unified interface in TypeScript and Python for every tool
- Automatic OAuth — Handles each integration's auth flow and token refresh
- Model agnostic — Works with OpenAI, Anthropic, and self-hosted LLMs through the same tool layer
- Semantic tool retrieval — Smart selection of the most relevant tools to avoid context overflow
- Observability — Built-in tool-call logs and failure replay
Use Cases
💡 Give any LLM access to 100+ tools (Slack, GitHub, Notion) in a few lines of code
💡 Expose internal SaaS tools to AI agents through a single Metorial layer
💡 Connect ChatGPT/Claude to user-authorized third-party services via automatic OAuth
💡 Run multi-model A/B tests against the same tool set to compare tool-use quality
💡 Use semantic tool retrieval to avoid stuffing long tool definitions into prompts
Tags
Categories
Quick Start
# Install the SDK
npm install metorial
# or Python: pip install metorial
# Initialize and run a session
import { Metorial } from 'metorial';
const metorial = new Metorial({ apiKey: process.env.METORIAL_API_KEY });
const session = await metorial.sessions.create({
deploymentId: 'your-deployment-id',
provider: 'openai',
model: 'gpt-4o',
tools: ['slack', 'github', 'notion', 'gmail'],
});
const result = await session.run({ prompt: 'Schedule a standup for tomorrow' });