@letretro/mcp

Your AI coding tools,
now retro-native

Read boards, search cards, create polls, generate reports — all from Claude, Cursor, VS Code, or any MCP client. One command, zero config, instant setup.

npmGitHubMCP Protocol
terminal
$npx @letretro/mcp
$npx @letretro/mcp install

> LetRetro MCP Server v1.0.0

> 10 clients detected

> ready

Two ways to use it

Install once, use everywhere

One package, two modes. Set up in seconds, then let your AI do the work.

Install Mode

npx @letretro/mcp install

  • Prompts for your API key — masked input for security
  • Validates the key against the server instantly
  • Auto-detects 10 MCP clients on your machine
  • Writes the right config file for each client you pick

Server Mode

npx @letretro/mcp (default)

  • Runs as a local stdio server your AI talks to
  • Handles the full MCP protocol — tools, reads, notifications
  • Forwards every request securely to mcp.letretro.com
  • Shuts down cleanly when you're done
In action

See how simple it is

Type your key, pick your clients, and you're done. No manual config editing.

installer

$ npx @letretro/mcp install

> Enter your LetRetro API Key: ••••••

API key validated successfully

> Detected clients:

Claude DesktopCursorVS CodeOpenCodeWindsurfContinue

> Select clients to configure (space to toggle)

All 10 clients selected

Claude Desktop — Updated

Cursor — Created

VS Code — Created

... and 7 more configured

Restart your MCP clients to start using LetRetro tools.

Clients

Works with every major MCP client

10 clients, one setup. Install once, use everywhere.

Claude Desktop
claude_desktop_config.json
Cursor
~/.cursor/mcp.json
VS Code
VS Code
~/.vscode/mcp.json
OpenAI Codex CLI
OpenAI Codex CLI
~/.codex/config.toml
OpenCode
~/.config/opencode/opencode.json
Antigravity (Gemini)
Antigravity (Gemini)
~/.gemini/config/mcp_config.json
Hermes AI
Hermes AI
~/.hermes/config.yaml
OpenClaw
OpenClaw
~/.openclaw/openclaw.json
Windsurf
~/.codeium/windsurf/mcp_config.json
C
Continue
~/.continue/config.json
17 tools

What you can do from your AI

Every action you'd take in the dashboard — now right from your coding assistant.

Boards

list_boards

Browse every board you own or are a member of

get_board

View full board detail with columns, settings, and members

get_board_stats

Check card counts, poll activity, and who's active

create_board

Create a new retro board with your preferred column template

Cards

get_board_cards

Read cards grouped by column with author profiles and comments

search_cards

Search across all card content in seconds

create_card

Drop a rich-text card into any column (up to 10K characters)

add_comment

Jump into conversations with @mentions

Polls & Reports

list_polls

See all polls with vote counts and whether they're still open

get_poll

Drill into poll results — who voted and how

create_poll

Run a poll with 2-20 options and priority labels

get_job_status

Check if an async job (report, export) is done yet

generate_report

Ask your AI to generate a retro trend analysis or summary

Team & Account

list_teams

See teams you're in and your role in each

get_activity_summary

Get a birds-eye view of recent cards, comments, and polls

get_profile

View your profile, subscription tier, and plan details

get_usage

Check your MCP API usage and remaining quota

Why LetRetro MCP

The simplest way to bridge AI and retros

One command setup

A single npx command. No JSON files, no config hunting.

10 clients supported

Claude, Cursor, VS Code, Windsurf — every major MCP client works.

Zero manual config

Auto-detects clients on your machine and writes the right config.

17 tools at your fingertips

Boards, cards, polls, teams, reports — control everything from your AI.

Faster, low latency

Edge-deployed server means every tool call returns in milliseconds.

Get started

Up and running in 30 seconds

01

Get your API key

Sign up for a Pro plan and generate a lr_-prefixed API key from your dashboard.

Go to dashboard
02

Run the installer

Run npx @letretro/mcp install — enter your API key, pick your clients, and the CLI writes all the config.

03

Restart and go

Restart your MCP clients. Your AI can now read boards, search cards, create polls, and more.

Connect with Claude

Claude Desktop

The installer auto-detects Claude Desktop and updates its config file:

claude_desktop_config.json

Claude Code

After running the installer, Claude Code can use LetRetro tools directly in the terminal.

npx @letretro/mcp install && npx @letretro/mcp

Every other MCP client

Cursor, VS Code, Windsurf, OpenCode, and 6 more — all auto-detected and configured. One install, all clients.

Try it yourself

What to ask your AI

Copy these prompts once LetRetro MCP is connected.

Cards

Show me the latest 5 cards from our retro board

FAQ

Frequently asked questions

Everything you need to know about LetRetro MCP.

MCP (Model Context Protocol) is an open standard that lets AI coding assistants communicate directly with tools and services. Instead of manually copying data between apps, your AI can read boards, search cards, and create polls in real time. It's like giving your AI a direct line to LetRetro.

Any assistant that supports the MCP protocol: Claude Desktop, Claude Code, Cursor, VS Code (via GitHub Copilot), OpenAI Codex CLI, Windsurf, OpenCode, Antigravity (Gemini), Hermes AI, OpenClaw, Continue, and more. We auto-detect 10 clients on your machine.

Sign up for a Pro plan on the letRetro dashboard. Once subscribed, generate a lr_-prefixed API key from the billing settings page. Then run npx @letretro/mcp install and paste the key when prompted.

Run npx @letretro/mcp install in your terminal. The interactive wizard will ask for your API key, auto-detect your MCP clients, and write the correct config files. No manual JSON editing required.

Your API key is stored locally in the MCP client config files on your machine (e.g., claude_desktop_config.json). It is sent securely to mcp.letretro.com with every request and is never logged or stored on our servers beyond the request lifecycle.

MCP access is included with the Pro plan. You get 20,000 requests per month. Check our pricing page for plan details and upgrade options.

Yes. You can manually add the MCP server config to any supported client's config file. The server entry points to npx @letretro/mcp with your API key set as the LETRETRO_API_KEY environment variable.

The server returns a quota exceeded response. Your usage resets at the start of each billing cycle. You can check your remaining quota anytime by calling the get_usage tool from your AI assistant.

Still have questions?

Reach out — we are happy to help.

Contact us