Weavz

Getting Started

Weavz is an integration infrastructure platform (embedded iPaaS) that provides connection management, MCP servers, triggers, and action execution for third-party services. It is designed to let you ship integrations in minutes rather than weeks, whether you are building an AI agent, a SaaS product, or an internal tool.

Why Weavz

Modern applications need to talk to dozens of external services. Managing OAuth flows, storing credentials securely, normalizing API schemas, handling webhooks, and keeping everything running across different runtimes is tedious and error-prone. Weavz handles all of that behind a single REST API and an MCP-compatible protocol layer.

  • 31 integrations out of the box, including Slack, GitHub, Google Sheets, Notion, Airtable, HubSpot, Stripe, Discord, Linear, Jira, and more.
  • Global edge delivery on Cloudflare Workers with auto-scaling and zero cold starts.
  • MCP-first approach: every integration is exposed as an MCP tool, making it trivial to connect AI agents to real-world services.
  • Private cloud deployments available for enterprise customers. Contact sales for details.

Key Concepts

Before diving in, it helps to understand the five building blocks that make up the Weavz platform.

Pieces

A piece is a packaged integration for a specific service (e.g. slack, github, google-sheets). Each piece declares its authentication requirements, available actions, and available triggers. Weavz ships with 31 pieces and the list grows with every release.

Connections

A connection stores the encrypted credentials needed to call a third-party API. Weavz supports OAuth2 (with PKCE), API keys, and custom auth schemes. Connections are scoped to an organization and optionally to a project, and they use AES-256 encryption at rest.

Learn more about connections

MCP Servers

An MCP server exposes a set of pieces as tools via the Model Context Protocol. You create a server, select which pieces it should offer, and Weavz generates a bearer token and an SSE endpoint that any MCP-compatible client (Claude, Cursor, custom agents) can connect to.

Servers operate in one of two modes: TOOLS mode (one tool per action, full JSON Schema for each) or CODE mode (three meta-tools that reduce context usage by 80–98%).

Learn more about MCP servers

Triggers

A trigger listens for events in a third-party service and delivers them to your application. Weavz supports both webhook-based triggers (real-time) and polling-based triggers (interval-based). Triggers include built-in deduplication so your handlers only see each event once.

Learn more about triggers

Actions

An action is a single operation you can execute against a third-party service, such as “send a Slack message”, “create a GitHub issue”, or “add a row to Google Sheets”. Actions are defined by pieces and invoked through the REST API, MCP tools, or the dashboard Playground.

Architecture at a Glance

Weavz is structured as a monorepo with two main packages:

  • api/ — The backend, built with Hono. Runs on both Cloudflare Workers and Node.js. Handles authentication, connection management, piece loading, action execution, trigger processing, and MCP protocol.
  • dashboard/ — The web UI, built with React 19, Vite, TanStack Router, and shadcn/ui. Provides a complete management interface for all platform resources.

The API uses a provider abstraction layer so that all business logic is runtime-agnostic. Database operations go through Drizzle ORM with PostgreSQL. Caching uses Redis (Node.js) or KV (Workers). Object storage uses S3/MinIO (Node.js) or R2 (Workers).

Authentication

Weavz supports four authentication paths:

  1. API key — For SaaS customers. Pass Authorization: Bearer wvz_....
  2. Session cookie — For dashboard users via BetterAuth.
  3. MCP bearer token — For MCP clients. Pass Authorization: Bearer mcp_....

Next Steps

Ready to get started? Here is the recommended path:

  1. Installation — Set up Weavz locally or sign up for the managed platform.
  2. Quick Start — Create your first connection and execute an action in under five minutes.
  3. MCP Servers — Connect an AI agent to your integrations.
  4. Code Mode — Learn how CODE mode reduces context usage for AI agents.