ConnectWise Automate MCP Server

An MCP server for the ConnectWise Automate REST API, the RMM that MSPs use to monitor and manage client endpoints. It lets an AI agent read computers, clients and their locations, alerts, the script catalog and script history and, on explicit request, reboot computers, run scripts and catalog commands, and create or update clients.

• Transport: MCP Streamable HTTP at POST /mcp, health probe at GET /health

• Stateless and multi-tenant: credentials arrive on every request as HTTP headers and are never stored; the API token is requested per request and discarded with it

• Python 3.12, uv, FastMCP, httpx

• License: Apache-2.0, Copyright 2026 MSPbots.ai

MSPbots Connector

The MCP Gateway built for MSPs: visible, controlled, and your data stays where it belongs.

Get started at connect.mspbots.ai →

MSPbots Connector gives your AI clients one MCP URL to reach your entire MSP stack (RMM, PSA, documentation, security, finance), and gives you control over who can call which tool and a record of everything that was called.

Hosted service: connect.mspbots.ai · Free to use.

Why Connector

Quick start

1. Sign up at connect.mspbots.ai.

2. Connect your vendors. Pick from 150+ connectors and authorize each one on the vendor's own page, or bring your own API keys. No extra vendor accounts needed.

3. Set access. Create roles and decide which tools each role may call. Nothing is allowed until you allow it.

4. Add your MCP URL to your AI client. Copy the URL from your Connector dashboard into Claude, ChatGPT, Cursor, VS Code, Gemini, Codex, or any MCP-capable client.

5. Ask your AI to do real work, and watch every call appear in the audit log.

Features

Unified gateway

• One MCP URL for all your AI clients

• 150+ connectors across the MSP stack

• Custom MCP servers supported, free

• Authorization happens on each vendor's own page; the AI never sees passwords or tokens

• Bring your own credentials: use the vendor API keys you already have

• Multiple accounts per app; connect once, reuse across AI clients

• Tool search: finds the right action for the task

• One sign-in; vendor credentials are injected behind the scenes and OAuth tokens refresh automatically

Visibility

• Audit log for every tool call (who, which vendor, when)

• Usage dashboard: call volume, top vendors, error rate, per-user activity

• Export audit logs to CSV, or stream to Loki, Graylog, or LogScale (SIEM)

• Optional recording of tool arguments and result summaries (off by default, enabled per organization)

Control

• Per-role, per-tool allowlist, deny by default

• owner / admin / member roles

• SCIM provisioning with Entra ID, Okta, JumpCloud, and Google Workspace

Security

• Credentials encrypted one record at a time (AES-256-GCM, per-record key)

• AI clients authenticate with OAuth 2.1 + PKCE

• Vendor responses are scanned for prompt-injection patterns and returned as structured data. This is one layer of defense, not a guarantee against every attack.

• Static egress IP for customer firewall and API allowlisting

Pricing

Free.

Links: connect.mspbots.ai · mspbots.ai

Related MCP server: ConnectWise Manage MCP Server

Authentication headers

Where to get them:

• Client ID: the API client id ConnectWise issues to integrations through its developer network (developer.connectwise.com). It identifies the integration, not the MSP.

• Integrator user: create an Integrator Login in the Automate system dashboard's integration settings (the exact menu differs between Automate versions; see the ConnectWise documentation for yours), and grant the permissions the agent needs (computers, clients, scripts, commands). Integrator logins have no 2FA, which makes them the right choice for a server-to-server connection.

• User auth (X-Cwa-Auth-Method: user) signs in as a normal Automate user. With 2FA on, a passcode is only valid for a short time, so this mode suits interactive testing rather than a stored connection.

• The API user's permissions apply: command levels and group "Send Commands" rights decide which commands and scripts may run, and client access decides which records are visible.

How credentials are sent to Automate: the server calls POST {X-Cwa-Server}/cwa/api/v1/apitoken with the ClientId header and the user name and password, then sends every API call with Authorization: Bearer <token> and the ClientId header. A 401 gets a fresh token once. Values that look like unresolved templates (${...}) are treated as missing.

A request to /mcp without the four required headers gets HTTP 401 naming the missing ones; a non-https server URL or an unknown auth method does not block tools/list; tool calls then return an invalid_argument error and the server URL is never called.

Tools

All tools return a JSON string. List tools take limit (default 50, max 200) and skip (a multiple of limit; Automate pages by page number) and return the records plus {count, skip, has_more, next_skip}.

Errors come back as a JSON envelope instead of an exception:

{"error": {"code": "unauthorized", "message": "Automate rejected the username / password or client id (token request failed)", "retryable": false}}

code is one of not_configured, unauthorized, not_found, invalid_argument, rate_limited, upstream_error. Reads are retried with backoff on 429, 5xx and network errors; writes are retried only on 429. When a write loses its connection before Automate answers, the result is {"completed": false, "interrupted": true, ...} instead of an error, because the action may already have run.

Run locally

uv sync
uv run python -m connectwise_automate_mcp          # listens on 0.0.0.0:8080

or with Docker:

docker build --platform linux/amd64 -t connectwise-automate-mcp:dev .
docker run --rm -p 8080:8080 connectwise-automate-mcp:dev

Configuration (environment variables, all optional): MCP_HTTP_HOST (default 0.0.0.0), MCP_HTTP_PORT (default 8080), LOG_LEVEL (default INFO). Credentials are never read from the environment.

Testing example

# list tools
curl -s http://localhost:8080/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -H "X-Cwa-Server: https://your-server.hostedrmm.com" \
  -H "X-Cwa-Client-Id: your-client-id" \
  -H "X-Cwa-Username: your-username" \
  -H "X-Cwa-Password: your-password" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}'

# offline computers of one client
curl -s http://localhost:8080/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -H "X-Cwa-Server: https://your-server.hostedrmm.com" \
  -H "X-Cwa-Client-Id: your-client-id" \
  -H "X-Cwa-Username: your-username" \
  -H "X-Cwa-Password: your-password" \
  -d '{"jsonrpc":"2.0","id":2,"method":"tools/call","params":{"name":"cwautomate_computers_list","arguments":{"client_id":12,"status":"offline"}}}'

Unit tests (no network access needed):

uv run pytest

Delivery notes

• Tool count (20). Kept as is because the tool names are an existing contract; every tool is listed from the first tools/list.

• Endpoint actions. cwautomate_computers_reboot, cwautomate_computers_run_script, cwautomate_computers_run_command and cwautomate_scripts_execute act on live machines and are annotated destructive; cwautomate_clients_create / _update change records. Their descriptions tell the agent to use them only on explicit request. There are no delete tools.

• Waiting for results. Automate has no synchronous run. Commands are polled through their execution row and scripts through new entries in the computer's script history, every 3 s for up to 90 s by default (timeout_seconds, at most 100 s so the call finishes inside the gateway's limit). An unfinished run returns completed: false with ids to check later.

• No interactive prompts. Older versions asked the user for a client filter before an unfiltered list, which a stateless gateway cannot relay; lists now page instead.

• Result size exception (SOP 2.5). cwautomate_computers_search and cwautomate_scripts_history have no paging (no skip), so they keep a maximum of 1000; results are still cut to 20,000 characters with a truncation notice. The paged lists take at most 200 per call.

• Alerts are read-only in the Automate API (no acknowledge / close).

License

Apache License 2.0 - see LICENSE and NOTICE.