Connect Lampwick to your AI agent

The MCP server at https://app.lampwick.dev/mcp lets an AI agent read your issues, requests, logs and monitors and act on them. Every tool call is an API request with your personal token, so the agent sees what you see.

Requirements

• A personal API token, created under API tokens on Account & security. It starts with lwp_.
• An MCP client that supports the Streamable HTTP transport and a custom Authorization header.

The examples read the token from the LAMPWICK_TOKEN environment variable:

export LAMPWICK_TOKEN=YOUR_LAMPWICK_TOKEN

YOUR_LAMPWICK_TOKEN is your lwp_… token. Where a client can't read environment variables, paste it in place of YOUR_LAMPWICK_TOKEN.

Add Lampwick to your client

Jump to: Claude Code · Codex CLI · Cursor · VS Code · Gemini CLI · Zed · Cline · Claude API · OpenAI and xAI · Other clients

Claude Code

claude mcp add --transport http lampwick https://app.lampwick.dev/mcp --header "Authorization: Bearer $LAMPWICK_TOKEN"

Or share it with your team in the project's .mcp.json, where Claude Code fills in ${LAMPWICK_TOKEN} from each person's environment:

{
  "mcpServers": {
    "lampwick": {
      "type": "http",
      "url": "https://app.lampwick.dev/mcp",
      "headers": { "Authorization": "Bearer ${LAMPWICK_TOKEN}" }
    }
  }
}

Codex CLI

In ~/.codex/config.toml:

[mcp_servers.lampwick]
url = "https://app.lampwick.dev/mcp"
bearer_token_env_var = "LAMPWICK_TOKEN"

Cursor

In ~/.cursor/mcp.json:

{
  "mcpServers": {
    "lampwick": {
      "url": "https://app.lampwick.dev/mcp",
      "headers": { "Authorization": "Bearer ${env:LAMPWICK_TOKEN}" }
    }
  }
}

VS Code

In .vscode/mcp.json. VS Code asks for the token once and stores it:

{
  "inputs": [
    { "type": "promptString", "id": "lampwick-token", "description": "lampwick API token", "password": true }
  ],
  "servers": {
    "lampwick": {
      "type": "http",
      "url": "https://app.lampwick.dev/mcp",
      "headers": { "Authorization": "Bearer ${input:lampwick-token}" }
    }
  }
}

Gemini CLI

gemini mcp add --transport http --header "Authorization: Bearer $LAMPWICK_TOKEN" lampwick https://app.lampwick.dev/mcp

Zed

In your Zed settings:

{
  "context_servers": {
    "lampwick": {
      "url": "https://app.lampwick.dev/mcp",
      "headers": { "Authorization": "Bearer YOUR_LAMPWICK_TOKEN" }
    }
  }
}

Cline

In Cline's MCP settings:

{
  "mcpServers": {
    "lampwick": {
      "type": "streamableHttp",
      "url": "https://app.lampwick.dev/mcp",
      "headers": { "Authorization": "Bearer YOUR_LAMPWICK_TOKEN" }
    }
  }
}

Claude API

With the MCP connector of the Messages API, send the beta header anthropic-beta: mcp-client-2026-09-15 and add to the request:

{
  "mcp_servers": [
    { "type": "url", "url": "https://app.lampwick.dev/mcp", "name": "lampwick", "authorization_token": "YOUR_LAMPWICK_TOKEN" }
  ],
  "tools": [
    { "type": "mcp_toolset", "mcp_server_name": "lampwick" }
  ]
}

OpenAI Responses API and xAI

Add Lampwick as a tool:

{
  "type": "mcp",
  "server_label": "lampwick",
  "server_url": "https://app.lampwick.dev/mcp",
  "authorization": "YOUR_LAMPWICK_TOKEN",
  "require_approval": "always"
}

For xAI, leave out require_approval.

Other clients

In other clients with a settings screen for MCP servers, add the URL https://app.lampwick.dev/mcp with the header Authorization: Bearer YOUR_LAMPWICK_TOKEN.

Check that it works

Ask your agent "Which Lampwick projects can you see?". It calls listProjects and lists the projects of your teams.

What you can 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?"
• "Which scheduled tasks are failing or missed their last run?"
• "Show the error logs from the last hour that mention timeout, then resolve issue #42."

What the agent can do

• Each API operation is a tool named after its operationId (listIssues, getIssue, updateIssue, …), with its path, query and body fields as arguments. Validation, access and owner rules are those of the API; an error comes back with isError: true.
• Read tools carry readOnlyHint; deletes and rotateProjectToken carry destructiveHint, so clients that honor the hints ask before running them.
• Not tools: token management, logging out, push devices, team invites, members and roles, and creating, changing or deleting status pages and notices.
• When the client connects, Lampwick tells the agent its conventions: durations in microseconds, the range values, id versus ref, and where to start.

Protocol details

Troubleshooting