Server catalog

The 28 bundled servers, what each one is for, and the environment variable it needs.

The current release ships 28 servers. This is the authoritative list from umt list; run it in your build to confirm.

Collaboration

Server	What it does	Required env
GitHub	Repository search, pull requests, workflows, issue triage	`GITHUB_TOKEN`
Notion	Search pages and databases, read docs, publish notes	`NOTION_TOKEN`
Slack	Channels, threads, and workspace updates	`SLACK_BOT_TOKEN`
Linear	Issue search, workflow state, new work items	`LINEAR_API_KEY`
Jira	Issues, tickets, incident triage	`JIRA_BASE_URL`, `JIRA_EMAIL`, `JIRA_API_TOKEN`
Discord	Guild discovery, channels, message history, messaging	`DISCORD_BOT_TOKEN`
Trello	Board and list discovery, card CRUD, archiving	`TRELLO_API_KEY`, `TRELLO_TOKEN`
Notion (MCP)	Full workspace integration with page and database CRUD	`NOTION_API_KEY`
Slack (MCP)	Channels, messages, users, and files	`SLACK_BOT_TOKEN`

Productivity

Server	What it does	Required env
Google Calendar	Calendars, events, meeting scheduling	`GOOGLE_CALENDAR_ACCESS_TOKEN`
Google Drive	Search, metadata, file export	`GOOGLE_DRIVE_ACCESS_TOKEN`

Data

Server	What it does	Required env
PostgreSQL	Schema inspection and guarded SQL	`POSTGRESQL_URL`
MongoDB	Collections and filtered document queries	`MONGODB_URI`
Redis	Keys, TTLs, cache diagnostics	`REDIS_URL`
Supabase	Tables, storage, project metadata	`SUPABASE_URL`, `SUPABASE_KEY`
Airtable	Table listing, record CRUD, filtering	`AIRTABLE_API_KEY`, `AIRTABLE_BASE_ID`

Platform

Server	What it does	Required env
Vercel	Projects, deployments, environment settings	`VERCEL_TOKEN`
Cloudflare Workers	Workers, routes, edge rollout state	`CLOUDFLARE_API_TOKEN`, `CLOUDFLARE_ACCOUNT_ID`
Docker	Containers, images, daemon state	none
npm Registry	Package search, versions, release metadata	none

Research and media

Server	What it does	Required env
Hacker News	Top stories, search, discussion threads	none
arXiv	Paper search and literature digests (experimental)	none
Spotify	Search tracks, playback, playlists	`SPOTIFY_ACCESS_TOKEN`
Stripe	Customers, invoices, subscriptions	`STRIPE_SECRET_KEY`

Local, memory, automation, AI

Server	What it does	Required env
FileSystem	Read and write inside explicitly allowed roots	`FILESYSTEM_ROOTS`
MemOS	Local-first persistent memory over MCP	none
Playwright (MCP)	Browser automation and scraping	none
OpenAI (MCP)	Chat, completions, embeddings	`OPENAI_API_KEY`

Safety knobs

Several servers expose write controls. Keep them off unless you need writes:

Variable	Effect
`POSTGRESQL_ALLOW_WRITES`	Enable write queries
`REDIS_ALLOW_WRITES`	Enable key mutation
`MONGODB_ALLOW_WRITE_PIPELINES`	Enable aggregation writes
`FILESYSTEM_ROOTS`	Sandbox the filesystem server
`FILESYSTEM_MAX_READ_BYTES`	Cap read size
`VERCEL_TEAM_ID`	Scope Vercel access to a team

The root .env.example has the full set.

Experimental and opt-in

umt list marks experimental servers explicitly; they may change or be removed. Third-party MCP servers can be added as opt-in entries using the same mcpServers shape. They are not part of the default bundle. One example is Vynly (@vynly/mcp), a social network for AI-generated images and short video with a free demo token on first call.

Verify

umt list
umt search postgres
umt test github        # live MCP handshake
umt conformance        # registry config + local handshakes

If a server is missing, check whether it is disabled in this build and whether its package is installed in workspace mode.

PreviousUMT overview Next CLI reference

Edit this page on GitHub