Chainitor MCP server: blockchain alerts from AI agents

Ask Claude, ChatGPT, Cursor or VS Code to set up, check and pause your on-chain alerts. One URL, OAuth sign-in, on every plan.

Chainitor runs a remote Model Context Protocol (MCP) server. Point an MCP client at it, sign in, and your agent can do what you would otherwise do in the Chainitor app: create a wallet balance or contract event monitor, find which monitors are unhealthy, read recent alerts, add an alert channel, or pause a noisy monitor.

Server URL: https://mcp.chainitor.com/mcp
Transport: Streamable HTTP
Sign-in: OAuth 2.1 with PKCE and dynamic client registration; you approve each connection on app.chainitor.com
Plans: Every Chainitor plan, including the 7-day free trial. Custom connectors in Claude.ai and ChatGPT depend on your plan with those apps.

Connect your AI client

Every client uses the same URL. The first time it connects, your browser opens a Chainitor consent page where you choose the organization and the permissions to grant.

Claude Code

Run this in a terminal:

claude mcp add --transport http chainitor https://mcp.chainitor.com/mcp

Start Claude Code, type /mcp, choose chainitor and select Authenticate. Your browser opens app.chainitor.com to approve the connection.

Claude.ai and Claude Desktop

Open Settings → Connectors → Add custom connector and enter https://mcp.chainitor.com/mcp as the URL.
Connect, sign in to Chainitor and approve the permissions you want to grant.

ChatGPT

Open Settings → Apps & Connectors → Advanced and turn on developer mode.
Select Create, enter https://mcp.chainitor.com/mcp as the server URL with OAuth as the authentication, then sign in to Chainitor.

Cursor

Add the server to ~/.cursor/mcp.json, or to .cursor/mcp.json in a project:

{
  "mcpServers": {
    "chainitor": {
      "url": "https://mcp.chainitor.com/mcp"
    }
  }
}

Open Cursor Settings → MCP and log in to chainitor when Cursor asks, to sign in to Chainitor.

VS Code (GitHub Copilot)

Run Chat: Open Customizations from the command palette, select MCP Servers, then Add Server.
Choose HTTP, enter https://mcp.chainitor.com/mcp, name it chainitor, and sign in to Chainitor when VS Code asks.

MCP Inspector

Run the inspector:
```
npx @modelcontextprotocol/inspector
```
Set the transport to Streamable HTTP and the URL to https://mcp.chainitor.com/mcp, then connect and complete the OAuth sign-in.

Things to ask your agent

“What Chainitor account am I connected to and how many monitors do I have left?”
“List my monitors and tell me which are unhealthy.”
“Alert me at [email protected] when 0xd8dA6BF26964aF9D7eEd9e03E53415D37aA96045 drops below 4 ETH on Ethereum.”
“Pause that monitor.”

When an agent sets up email alerts to an address Chainitor has not seen before, that address first receives a verification email. Alerts reach it once the link in the email is clicked.

Tools the server exposes

An agent only sees the tools its connection's scopes allow. Plan limits apply exactly as they do in the app: an agent cannot create more monitors, or check more often, than your plan permits.

Always available

Scopes: any connection

get_account: The connected user and organization, its plan and monitor usage, and what this connection is allowed to do.
list_monitor_types: Every monitor type with its configuration fields, whether your plan allows it and its minimum check interval, plus the supported chains.

Read

Scopes: monitors:read, channels:read, alerts:read

list_monitors: Your monitors, newest first, with their latest status.
get_monitor: One monitor with its configuration, channels and latest result, plus its five most recent alerts when alerts:read is granted.
list_alerts: Recent alerts across your monitors, newest first.
list_channels: Your alert channels. Channel secrets are never returned.

Manage monitors

Scopes: monitors:write

create_monitor: Create a monitor of any type your plan allows, optionally with email alerts in the same call (a new email address also needs channels:write).
update_monitor: Rename a monitor, change its check interval or configuration, or replace its alert channels.
pause_monitor: Stop checking a monitor and sending its alerts until it is resumed.
resume_monitor: Start checking a paused monitor again.
run_monitor_now: Check a monitor immediately instead of waiting for its next run. Alerts fire as on a scheduled run.

Manage alert channels

Scopes: channels:write

create_channel: Add an email, Slack or webhook alert channel.
send_test_notification: Send a test alert to a channel to prove it is wired up.

Security and control

OAuth, not shared keys. Clients sign in with OAuth 2.1 and PKCE, and you approve each connection on app.chainitor.com. No Chainitor password or API key is pasted into the AI client.
Scoped access. Your AI client requests some of monitors:read, monitors:write, channels:read, channels:write and alerts:read; the consent page shows what it asked for, and its Read-only access switch drops the write scopes. A connection only sees the tools its scopes allow.
One user, one organization. Each connection acts as you, inside the one organization you chose when connecting.
No delete tool. Agents pause monitors instead; deleting stays a deliberate action in the app.
Audit log. Changes an agent makes are recorded in your organization's audit log, alongside changes made in the app.
Disconnect any time. Revoke an app under Connected apps on the API Keys page of the Chainitor app.
Still non-custodial. Chainitor reads public chain data and sends alerts. Neither the app nor the MCP server can sign or send a transaction.

Prefer code? The REST API

Scripts and backends can manage monitors over plain HTTP. The REST API lives at https://app.chainitor.com/api/v1 and authenticates with an API key you create on the API Keys page of the Chainitor app. Like the MCP server, it is on every plan, including the trial.

OpenAPI 3.1 specification: load it into a client generator, Postman or an agent framework.
REST API reference: every endpoint, parameter and response.
Interactive API explorer (Swagger UI in the app).

Frequently asked questions

What is the Chainitor MCP server?

A remote Model Context Protocol server at https://mcp.chainitor.com/mcp. It lets an AI agent such as Claude, ChatGPT, Cursor or VS Code read and manage your Chainitor monitors, alerts and alert channels on your behalf, within the same plan limits as the Chainitor app.

Which plans include the MCP server and the REST API?

Every plan, including the 7-day free trial. Both work for any account with an active subscription or trial; your plan only sets the limits that already apply in the app, such as how many monitors you can create and how often they run.

Can an AI agent delete my monitors?

No. There is no delete tool, by design. An agent can pause a monitor, which stops it and can be undone; deleting stays something you do yourself in the Chainitor app.

How do I limit what an agent can do?

Turn on Read-only access on the Chainitor consent page when you connect. A read-only connection never sees the write tools. Each connection covers one user and one organization, and you can revoke it at any time.

How do I disconnect an AI app?

In the Chainitor app, open the API Keys page and revoke the app under Connected apps.

Can I see what an agent changed?

Yes. Changes made through the MCP server, such as creating, updating, pausing or running a monitor and creating or testing a channel, are recorded in your organization's audit log.

Let your agent watch the chain

Start a free trial, connect your AI client, and ask it to set up your first monitor.

Start monitoring free

7-day free trial • No immediate charge