# Connect Lampwick to Claude

> Connect Lampwick from Claude's connector directory, sign in and choose read-only or read & write. Works in claude.ai, Desktop, mobile and Claude Code.

With Lampwick connected, Claude can read your issues, requests, logs and uptime monitors, and act on them if you allow it. You sign in to Lampwick once and pick the access; there is no token to copy. The connector works in claude.ai, Claude Desktop, the Claude mobile apps and Claude Code.

You need a Lampwick account. On Claude's Team and Enterprise plans an Owner adds connectors for the organization (see [Team and Enterprise](#team-and-enterprise)).

## Connect from the directory

1. Open [Lampwick in Claude's connector directory](https://claude.ai/directory/lampwick), or search for Lampwick under **Customize** → **Connectors** → **Browse connectors**.
2. Click **Connect**. Lampwick opens in your browser: log in, with your two-factor code if you turned it on.
3. On **Connect … to Lampwick?**, choose **Read-only (recommended)** or **Read & write**, then click **Connect**. Pick **Read-only** unless you want Claude to change things: it is the safe default, and you can connect again with **Read & write** later.

## Or add it as a custom connector

The same server, added by hand (for example in a Claude organization that only allows custom connectors):

1. Open **Customize** → **Connectors**.
2. Select **Add**, then **Add custom connector**.
3. Enter the name `Lampwick` and the URL `https://app.lampwick.dev/mcp`, then click **Continue**.
4. Keep the authentication settings Claude marks **Detected**. Leave the OAuth client ID and secret empty: Lampwick accepts Claude's published identity and automatic registration, so you don't create a client. Click **Add**.
5. Click **Connect**. Lampwick opens in your browser: log in, with your two-factor code if you turned it on.
6. On **Connect … to Lampwick?**, choose **Read-only (recommended)** or **Read & write**, then click **Connect**. Pick **Read-only** unless you want Claude to change things: it is the safe default, and you can connect again with **Read & write** later.

Back in Claude, the connector shows **Connected** under **Your connectors**. Its page lists Lampwick's tools under **Tool permissions**.

### Team and Enterprise {#team-and-enterprise}

An Owner adds the connector once for the organization: **Organization settings** → **Connectors** → **Add** → **Custom** (type **Web** if Claude asks), with the same name and URL. Members then find Lampwick under **Customize** → **Connectors** with the **Custom** label, click **Connect** and sign in with their own Lampwick account. Each member sees only the Lampwick teams they belong to.

## Claude Desktop and the mobile apps

Claude Desktop and the Claude mobile apps use the connectors of your Claude account. Add Lampwick once as above, and it is there in each app. Sign-in goes through the same Lampwick consent screen.

## Claude Code

Add the server without a token, then run `/mcp` in Claude Code and sign in from there:

```bash
claude mcp add --transport http lampwick https://app.lampwick.dev/mcp
```

Claude Code opens Lampwick in your browser for the same sign-in and access choice. To use a personal token instead, see [the Claude Code token setup](/docs/mcp/#claude-code).

## Check that it works

In a chat, click **+** → **Connectors** and turn Lampwick on. Then ask "Which Lampwick projects can you see?". Claude calls `listProjects` and lists the projects of your teams.

Other things to ask:

- "Which open issues in Acme Shop were first seen in the last 24 hours?"
- "What are the 10 slowest routes by p95 this week?"
- "Which uptime monitors are down right now?"

## Read-only or read & write

- **Read-only**: Claude looks at issues, requests, logs, scheduled tasks and monitors. It can't change anything.
- **Read & write**: Claude also resolves or ignores issues, adds and changes monitors, renames projects and deletes things. Never: delete a team, rotate a project's ingest token or change its browser key.

Every tool that changes something is marked as destructive, except the ones that only add something (a project, a monitor, a deploy marker). Claude asks you before it runs a destructive tool. You can also block single tools on the connector's page under **Tool permissions**.

What Claude reads includes error messages and logs from your apps, which could contain instructions meant for it. With read-only access, those can't change anything.

## Disconnect

- In Lampwick: **Account & security** → **Connected apps** → **Disconnect**. Claude's access stops at once.
- In Claude: **Customize** → **Connectors**, then **Remove** on Lampwick.

A connection that goes unused for 60 days stops working. Resetting your Lampwick password or turning two-factor authentication on or off disconnects every connected app; click **Connect** again to sign in.

## Troubleshooting

| What you see | Cause and fix |
|---|---|
| The connector shows **Connect** or **Reconnect** | Sign-in didn't finish, or the connection was disconnected. Click it and log in to Lampwick again. |
| Claude says the connection is read-only | You chose **Read-only**. Disconnect in Lampwick, connect again and choose **Read & write**. |
| A project is missing | Claude sees only the teams of the Lampwick account you signed in with. Ask an owner to invite you to the team. |
| A tool fails with `403` on a read & write connection | That action is for team owners. |

For other clients (ChatGPT, Cursor, VS Code, Codex) and for personal tokens, see [the MCP server docs](/docs/mcp/).