Weavz

Quick Start

This guide walks you through getting your first integration working end-to-end. By the end, you will have created a connection to a third-party service, executed an action through the Playground, and connected an AI agent via MCP.

Time required: about 5 minutes.

Prerequisites

  • A Weavz account on platform.weavz.io
  • A third-party account to connect (e.g. Slack, GitHub, or Google)

Step 1: Sign In and Create an Organization

Navigate to your Weavz dashboard. If this is your first time, create an account using email or a social provider. After signing in, you will be prompted to create an organization. An organization is the top-level container for all your resources — connections, MCP servers, triggers, and projects.

Give your organization a descriptive name (e.g. “My Company” or “Personal”). You can create additional organizations later.

Step 2: Create a Project (Optional)

Projects let you scope resources for different applications or environments. Navigate to Projects in the sidebar and create one. If you skip this step, resources will be scoped to the organization level.

Step 3: Create an MCP Server

Navigate to MCP Servers in the sidebar and click Create Server. Configure the server:

  1. Name: Give it a descriptive name, e.g. “Dev Tools”
  2. Mode: Select TOOLS for this quickstart (each action becomes a separate MCP tool)
  3. Pieces: Select the integrations you want to expose. For this guide, try slack and github.

After creation, the dashboard shows your server's MCP endpoint URL and bearer token (prefixed with mcp_). Copy both — you will need them in Step 6.

Step 4: Create a Connection

Navigate to Connections in the sidebar and click Create Connection.

  1. Select a piece (e.g. Slack)
  2. Choose auth type: For OAuth2 pieces like Slack, click Connect to launch the OAuth popup. You will be redirected to the service's authorization page. Grant access, and the popup will close automatically.
  3. For API key pieces (e.g. OpenAI), paste your API key into the form field.

Once created, the connection appears in your list with a Connected status. Weavz encrypts and stores the credentials using AES-256 encryption. OAuth2 tokens are automatically refreshed when they expire.

Step 5: Test via Playground

Navigate to Playground in the sidebar. The Playground provides a unified console for testing actions, triggers, MCP servers, and code snippets.

  1. Switch to the Actions tab
  2. Select a piece (e.g. slack)
  3. Select an action (e.g. send_message)
  4. The form renders dynamically based on the action's input schema. Fill in the required fields — for Slack, select a channel and type a message.
  5. Click Execute

The result appears in the output panel. If the action succeeded, you should see the API response from the third-party service (and, in this case, a message in your Slack channel).

Step 6: Connect an AI Agent via MCP

Now connect an MCP-compatible AI client to your server. Using the endpoint URL and bearer token from Step 3:

Claude Desktop

Add the following to your Claude Desktop MCP configuration (usually at ~/Library/Application Support/Claude/claude_desktop_config.json on macOS):

{
  "mcpServers": {
    "weavz": {
      "url": "https://your-instance.weavz.io/mcp/servers/YOUR_SERVER_ID/sse",
      "headers": {
        "Authorization": "Bearer mcp_YOUR_TOKEN_HERE"
      }
    }
  }
}

Cursor

In Cursor settings, add an MCP server with the SSE endpoint and bearer token authentication. Cursor supports MCP natively and will discover all available tools from your server.

Custom Agent

Any MCP client that supports SSE transport can connect. Point it at the server's SSE endpoint and pass the bearer token in the Authorization header.

Step 7: Execute via REST API

You can also execute actions programmatically using the REST API. Here is a curl example that sends a Slack message:

curl -X POST https://your-instance.weavz.io/api/v1/actions/execute \
  -H "Authorization: Bearer wvz_YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "pieceName": "slack",
    "actionName": "send_message",
    "input": {
      "channel": "#general",
      "text": "Hello from Weavz!"
    },
    "connectionId": "conn_abc123"
  }'

The response contains the action result along with execution metadata:

{
  "success": true,
  "output": {
    "ok": true,
    "channel": "C01ABC123",
    "ts": "1234567890.123456",
    "message": {
      "text": "Hello from Weavz!",
      "type": "message"
    }
  }
}

What to Explore Next

  • Connections — Learn about auth types, multi-tenant scoping, and connection policies.
  • MCP Servers — Understand TOOLS vs CODE mode and helper tool generation.
  • Code Mode — Reduce AI agent context usage by 80–98% with three meta-tools.
  • Triggers — Set up webhook and polling triggers to react to external events.
  • Connection Policies — Configure how connections are shared across users and projects.