MCP 服务器

findme-mcp

Enables AI clients to create and manage FindMe Photo wedding galleries, upload photos from local files or URLs, and retrieve analytics and QR codes using natural language.

README

FindMe Photo MCP Server

Official Model Context Protocol server for FindMe Photo — the AI-powered wedding gallery platform.

Drive your FindMe galleries from Claude, ChatGPT Codex, Cursor, and any other MCP-compatible AI client. Create events, upload photos, pull analytics — all in natural language.

What it does

"Create a FindMe event for the Sarah & Mike wedding on April 22, upload every photo in ~/Pictures/Sarah-Mike to it, then show me the QR code."

The AI uses this MCP server to do all three steps in one turn.

Tools

Tool	Purpose
`create_event`	Create a new gallery.
`list_events`	List your events (paginated).
`get_event`	Get event details + stats.
`update_event`	Rename, re-date, or change access code.
`delete_event`	Soft-delete with 7-day recovery.
`restore_event`	Un-delete within 7 days.
`upload_photos_from_paths`	Upload local files or folders. Primary upload tool.
`upload_photos_from_urls`	Upload from public URLs (Dropbox, direct Drive links, etc.).
`upload_photos_from_drive_folder`	Import from Google Drive (partial in v1).
`get_event_qr`	PNG QR code pointing to the public gallery.
`get_event_analytics`	Guest visits, unique visitors, downloads, selfie searches.
`get_usage`	Current-month API usage, storage, event totals.

Requirements

Node.js 18+
A FindMe Photo account on Free+, Growth, or Pro
A personal API key from https://findme.photo/settings/api

Install

The easiest way: let the AI install itself. Paste a prompt, the AI edits its own config. No hidden files.

For a step-by-step walkthrough aimed at non-technical photographers, see the full install guide.

Claude Desktop

Open Claude Desktop, click the Code tab (not Chat), then paste this prompt (replace the key):

Install the FindMe MCP server for me.

- Server name: findme
- Command: npx
- Args: -y findme-mcp
- Env var: FINDME_API_KEY = fm_live_PASTE_YOUR_KEY_HERE

Edit my claude_desktop_config.json under mcpServers, then tell me to fully quit and reopen Claude.

Claude Code (CLI)

claude mcp add findme --scope user --env FINDME_API_KEY=fm_live_xxxxx -- npx -y findme-mcp

ChatGPT Codex

codex mcp add findme --env FINDME_API_KEY=fm_live_xxxxx -- npx -y findme-mcp

Cursor / Windsurf / VS Code / Cline

Use the tool's built-in MCP install UI with:

Command: npx
Args: -y findme-mcp
Env: FINDME_API_KEY=fm_live_xxxxx

ChatGPT web / mobile

Coming soon — requires a hosted MCP endpoint. We're shipping mcp.findme.photo next. In the meantime, use Claude Desktop or Codex.

Any other MCP client

FINDME_API_KEY=fm_live_xxxxx npx findme-mcp

Verify

After restart, you should see findme in your client's tool list (12 tools). Try:

Create a FindMe event called "My First Gallery" for today, then show me the QR code.

Troubleshooting

"MCP server failed to start"

Ensure Node.js 18+: node --version
First run downloads the package via npx; allow it through your firewall
Check the JSON is valid if you edited manually (no trailing commas)

"Your FindMe API key is invalid or has been revoked"

Generate a fresh key at https://findme.photo/settings/api
Paste the full key including the fm_live_ prefix
Fully quit and restart your AI client

"API access requires Free+, Growth, or Pro"

API access is a paid-tier feature. Upgrade at https://findme.photo/pricing.

Upload fails with "can't find folder" on macOS

macOS privacy protection. Open System Settings → Privacy & Security → Full Disk Access and add your AI client (Claude Desktop, Terminal, etc.). Then quit and reopen.

Upload partially failed

The tool never errors a whole batch for one bad file — it returns a summary showing which files succeeded and which didn't. Ask the AI to retry the failed ones specifically.

Limits

Tier	Rate limit	Monthly quota
Free+	10 req/min	500 / month
Growth	60 req/min	10,000 / month
Pro	300 req/min	Unlimited

Per file: 50 MB photos, 500 MB videos, 50 files per upload batch.

v1 limitations

upload_photos_from_drive_folder requires an existing Drive import created via the findme.photo web UI; the MCP can poll status but not create imports yet.
The MCP runs locally on the same machine as your AI client; it only sees files on that machine.