Skip to main content
Along’s MCP server exposes your Safes as tools to any MCP-compatible AI assistant. Once connected, the assistant can list your Safes, select an active Safe, and query its knowledge graph — all from within the conversation.

​
MCP endpoint

Along provides two endpoint variants depending on how you want to scope access:
VariantEndpointDescription
User-scopedPOST /mcp/v1Access is tied to your user account; Safe selection happens at runtime
Safe-scopedPOST /mcp/v1/{safeId}Access is pinned to a specific Safe at the token level
For most integrations — including Claude and ChatGPT — you’ll use the user-scoped endpoint https://along-api-qrd5m37v3a-ey.a.run.app/mcp/v1) and select your Safe during the OAuth authorization flow or from within the conversation.

​
Authentication

All MCP requests require a Bearer token obtained through Along’s OAuth 2.0 PKCE flow. The token is issued after you authorize the connection in the Along consent page. 
POST /mcp/v1
Authorization: Bearer avo_...
Content-Type: application/json

​
Protocol

Along’s MCP server implements the MCP 2025-03-26 protocol using JSON-RPC 2.0 over HTTP. Requests are standard JSON-RPC method calls; responses are synchronous JSON objects.

​
Available tools

After connecting, the following tools are available to the AI assistant:
ToolDescription
along_list_safesLists all Safes your account has access to
along_select_safeSets the active Safe for the current session
alongQueries the active Safe’s knowledge graph

​
Get started

OAuth Flow

Understand how Along’s OAuth 2.0 PKCE authorization works and how tokens are issued.

Connect to Claude

Add Along as an MCP server in Claude and start querying your Safes from conversations.

Connect to ChatGPT

Add Along as a connector in ChatGPT and query your knowledge graph directly.

MCP Tools

Explore the full reference for Along’s MCP tools and their parameters.