IN CLAUDE

Bring your Nerve context
into Claude

Add Nerve as a custom connector in Claude, Claude Desktop, or Cowork, then sign in with your Nerve account. No API keys, no code.

GET STARTED →
MCP access starts on Pro
FOR DEVELOPERS

Connect Claude Code
or any MCP client

Use the MCP server with OAuth or an API key, or call the REST API for your briefing, agent status, action queue, and contacts.

SEE SETUP →
Read on Pro · read and write on Operator

What is MCP?

MCP (Model Context Protocol) is an open standard for connecting AI assistants to outside systems. Instead of pasting context into a chat window, an MCP client like Claude calls tools that read live data. Nerve's MCP server gives your client your daily briefing, your action queue, and your agents' status, and on Operator it can add actions and contacts and put events on your calendar.

Connect Nerve

Two ways in: a custom connector in Claude, or the Claude Code command line. Both sign in with your Nerve account, and Claude Code can use an API key instead.

OPTION A: CLAUDE, CLAUDE DESKTOP, OR COWORK

Add a custom connector
In Claude, go to Customize, then Connectors, click +, and choose Add custom connector. On a Claude Team or Enterprise plan, an owner adds it first under Organization settings, Connectors.
Enter the Nerve URL
Use https://getnerve.ai/mcp as the connector URL.
Sign in with your Nerve account
Claude runs the OAuth sign-in and refreshes the token, so there's no key to copy. Then ask "What's in my briefing?" or "What's waiting in my action queue?"

OPTION B: CLAUDE CODE

Add the server
Run this in your terminal:
Terminal
claude mcp add --transport http nerve https://getnerve.ai/mcp
Sign in
Run /mcp inside Claude Code and follow the sign-in steps in your browser.

Rather use an API key? Create one in Nerve under Settings, Developer and pass it as a header:

Terminal
claude mcp add --transport http nerve https://getnerve.ai/mcp --header "Authorization: Bearer nrv_YOUR_KEY"
.mcp.json at the project root, with NERVE_API_KEY set
{
  "mcpServers": {
    "nerve": {
      "type": "http",
      "url": "https://getnerve.ai/mcp",
      "headers": {
        "Authorization": "Bearer ${NERVE_API_KEY}"
      }
    }
  }
}

Read tools from Pro, write tools on Operator

Your client calls tools/list and gets the tools your plan includes. A few of them:

READ

nerve_briefing
Your current daily briefing: summary, priority actions, recommendations, and focus items.
nerve:read
nerve_list_actions
The actions waiting for your review, filterable by status, source, type, or lane.
nerve:read
nerve_agent_status
Health and last run time for each agent.
nerve:read

WRITE

nerve_add_action
Add an action to your queue for review in Nerve.
nerve:write
nerve_add_contact
Create a contact in Nerve, or update the one that matches.
nerve:write
nerve_find_time
Find open time on your connected calendar for a task.
nerve:write
nerve_calendar_create
Create an event on your connected calendar.
nerve:write

Two ways to authenticate

OAuth 2.1 for clients that sign in for you, or API keys for Claude Code, scripts, and direct REST calls. Both go in the Authorization header as Bearer tokens, and both are capped to what your plan includes.

OAuth 2.1
Claude and Claude Code run this flow for you. Access tokens work on the REST API too.
Dynamic client registration
PKCE with S256
1-hour access tokens, 30-day refresh
Scopes: nerve:read, nerve:write
API Keys
Create keys in Nerve under Settings, Developer, for Claude Code, scripts, and direct REST calls.
nrv_ prefix, shown once
A name and scopes per key
Revoke any key in Settings
Up to 10 new keys an hour
OAuth 2.1 endpoints
GET /.well-known/oauth-protected-resource
  → Protected resource metadata (RFC 9728)

GET /.well-known/oauth-authorization-server
  → Authorization server metadata (RFC 8414)

POST /oauth/register
  → Dynamic client registration (RFC 7591)

GET /oauth/authorize
  → Authorization, signing in with your Nerve account

POST /oauth/token
  → Authorization code and refresh token grants
Scopes
OAuth scopes
  nerve:read   → The read tools, and the REST endpoints the read key scope reaches
  nerve:write  → The read and write tools, and every REST endpoint listed here
  nerve:admin  → Accepted, with the same access as nerve:write

API key scopes
  read         → GET /api/mcp/briefing, GET /api/mcp/agents
  briefing     → GET /api/mcp/briefing
  agents       → GET /api/mcp/agents
  admin        → Every endpoint listed here, writes included

REST endpoints

Full schemas are in /openapi.json. The admin key scope covers every endpoint.

GET /api/mcp/briefing
  → Your latest briefing: summary, priority actions, recommendations, and focus items
  key scope: read or briefing · OAuth: nerve:read

GET /api/mcp/agents
  → Health, last run time, and summary for each agent
  key scope: read or agents · OAuth: nerve:read

GET /api/mcp/actions
  → Your action queue, filterable by status, source, type, and lane
  key scope: admin · OAuth: nerve:write

POST /api/mcp/actions
  → Add an action, or several at once as { actions: [...] }
  key scope: admin · OAuth: nerve:write

POST /api/mcp/contacts
  → Create a contact, or update the one with the same email or name
  key scope: admin · OAuth: nerve:write
Key management, with your signed-in Nerve session
GET /api/mcp/keys
  → List your keys and the scopes your plan allows

POST /api/mcp/keys
  → Create a key with a name and scopes. The raw key comes back once

DELETE /api/mcp/keys?hash=<hash>
  → Revoke a key

MCP and API access by plan

Lite
Free
  • No MCP server access
  • No API keys
Pro
$29/mo
  • ✓Read tools on the MCP server
  • ✓Read endpoints on the REST API
  • ✓Keys with the read, briefing, agents scopes
Operator
$79/mo
  • ✓Read and write tools on the MCP server
  • ✓Every REST endpoint on this page
  • ✓Keys with any scope, admin included

On team plans, Starter has no MCP access, Growth gets the read tools, and Enterprise gets read and write tools. See team plans.

Machine-readable files

Connect Nerve to your AI tools

MCP and API access starts on Pro. Pick a plan, then add the connector.

SEE PLANS →