whitelabel-fleet-mcp

whitelabel-fleet-mcp

Exposes file, shell, git, and memory tools from a host Mac to remote Claude clients (e.g., iPhone) via HTTPS and Tailscale, enabling full coding capabilities on the go.

Category
Visit Server

README

whitelabel-fleet-mcp

MCP server that exposes the whitelabel fleet to remote Claude clients. Plug your iPhone (or another laptop's Claude Desktop) into this and you get the same coding power you have on the host Mac: read/write files, run shell commands, drive git, read whitelabel-memory.

Why

Coding from a phone usually means SSH + a terminal app — passable but clunky, and Claude has no idea what the fleet looks like. Model Context Protocol fixes that: the iOS Claude app speaks MCP as a first-class client, so once this server is exposed, your phone's Claude can call list_dir, git_status, run_shell, etc., as native tools.

Architecture

iPhone Claude
    ↓ HTTPS (Streamable HTTP transport)
Tailscale Serve (or Funnel)
    ↓
127.0.0.1:8473  ← this server
    ↓
Fleet tools:
  • files       read_file, write_file, list_dir, search_code
  • shell       run_shell
  • git         git_status, git_diff, git_log, git_commit
  • memory      read_memory, append_journal

Auth: bearer token in the Authorization header. The token is the perimeter — guard it like a password.

Setup (local)

cd whitelabel-fleet-mcp
npm install
cp .env.example .env
echo "FLEET_MCP_TOKEN=$(openssl rand -hex 32)" >> .env
npm start

Verify:

curl http://127.0.0.1:8473/healthz
# → {"ok":true,"version":"0.1.0","sessions":0}

Full end-to-end smoke test (boots server, runs every tool, prints results):

FLEET_MCP_TOKEN=$(grep FLEET_MCP_TOKEN .env | cut -d= -f2) \
  node scripts/smoke.mjs

Setup (iPhone Claude)

  1. Install Tailscale on your iPhone and log in to the same tailnet as the host Mac.
  2. On the host Mac, expose the server over HTTPS:
    bash scripts/serve.sh
    
    This prints your tailnet hostname, e.g. https://mac1.tail1234.ts.net/mcp.
  3. In the iPhone Claude app:
    • Settings → ConnectorsAdd custom connector
    • URL: the tailnet /mcp URL from step 2
    • Auth: Bearer token → paste your FLEET_MCP_TOKEN
  4. Open a new chat. Claude will list the tools it picked up (read_file, git_status, run_shell, …). Try: "list the whitelabel-flow repo, then show me the most recent commits."

Setup (start at login)

bash scripts/install-launchd.sh

Installs dev.whitelabel.fleet-mcp as a user LaunchAgent that runs at login and restarts on crash. Logs at ~/Library/Logs/whitelabel-fleet-mcp/{stdout,stderr}.log.

Uninstall:

launchctl unload ~/Library/LaunchAgents/dev.whitelabel.fleet-mcp.plist
rm ~/Library/LaunchAgents/dev.whitelabel.fleet-mcp.plist

Tools

Tool What it does
read_file Read a UTF-8 file (up to 1 MB).
write_file Create or overwrite a file. Makes parent dirs.
list_dir List immediate children of a directory.
search_code rg (or grep -r) across the workspace.
run_shell Execute a shell command via zsh -lc. Capped at 600 s and 200 KB output.
git_status git status --short --branch for any repo.
git_diff Unstaged or staged diff, optionally narrowed to paths.
git_log Last N commits, oneline.
git_commit Stage + commit, with optional push.
read_memory Fetch whitelabel-memory/HANDOFF.md + most-recent journal.
append_journal Drop a new journal entry into whitelabel-memory.

All file/repo paths must resolve inside WORKSPACE_ROOT (set in .env, default ~/Documents). Paths outside are rejected.

Security notes

  • The bearer token is the only auth. Rotate it if you suspect leakage.
  • run_shell runs anything — gating is intentional only against Cloudflare/Tailscale-level mishaps, not the authenticated client. If you give the token to a friend they can do anything you can.
  • Tailscale Serve binds to your tailnet only; Funnel would expose it to the public internet (HTTPS-auth'd by token, not network ACL). Prefer Serve unless you need outside access.

Roadmap

Version Adds
v0.1 (here) core tools, bearer auth, Tailscale Serve, LaunchAgent
v0.2 dispatch_to_worker (call whitelabel-orchestrator API to fire long-running jobs on mac1-4)
v0.3 summon_claude_code (spawn a fleet claude -p for heavy refactors)
v0.4 per-tool ACLs (some clients get read-only, some can shell)
v1.0 replaces dispatch.sh entirely as the fleet's RPC layer

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured