MCP Servers in the Wild 🌴 👀 🌴 #2075
Replies: 20 comments 4 replies
|
Here's one of my favorites: the mBot MCP server Community member built an MCP server to control a bot with Goose
mbot.mp4 |
|
Found an MCP server that I would loooove to try out The ability to send emails directly from your MCP client 🤯 https://github.com/resend/mcp-send-email |
|
Not sure this thread is still used for the Goose MCP servers curation, but we just released a MCP Server with Kiwi.com, which allow you to search and book flights directly from your MCP Client: MCP Server Name: Kiwi.com Official Flight Search |
|
Thanks for sharing! Added it to our docs and made a little youtube short off of it |
|
Looks awesome, thanks! |
|
sorry to barge in on this thread but I've been unable to add this MCP endpoint for Kiwi, when I add the server to Goose it pops up a browser window with a 403 and this url:
the instructions are straightforward and curl from a terminal spits back this: 😬 any ideas? this is a macOS workstation (intel), but the result is the same on my Wintendo (windows11, intel) |
|
What options are you using in Goose? Looks like you are trying to use Oauth on the server? |
|
@blackgirlbytes I'd love to add TwelveLabs MCP Server into Goose official extensions directory! MCP Server Name: TwelveLabs MCP Server What it does
Key Features
|
|
Hi! I'd like to propose adding two Mapbox MCP servers to the Goose extensions list as I work at Mapbox on our MCP servers. Both solve different but complementary problems for AI agents. 1. Mapbox MCP Server (Location Intelligence)MCP Server Name: Mapbox MCP Server What It DoesGives AI agents comprehensive location intelligence capabilities:
Why This Is Valuable for GooseMost AI agents can't answer basic location questions like "How long does it take to drive from X to Y?" or "What's within walking distance of this address?" This server bridges that gap, enabling agents to:
Setup for GooseOption 1: Hosted (easiest) {
"mapbox": {
"url": "https://mcp.mapbox.com/mcp"
}
}First connection uses OAuth browser login to your Mapbox account (free tier available). Option 2: Local {
"mapbox": {
"command": "npx",
"args": ["-y", "@mapbox/mcp-server"],
"env": {
"MAPBOX_ACCESS_TOKEN": "your_token_here"
}
}
}Get a free access token at: https://account.mapbox.com/auth/signup/ Example Use Cases
Documentation2. Mapbox DevKit MCP Server (Map Style Designer)MCP Server Name: Mapbox DevKit MCP Server What It DoesA visual design assistant for creating and customizing Mapbox map styles:
Why This Is Valuable for GooseMap styling is inherently visual work. This server brings a unique interactive visual experience to Goose through MCP-UI support, allowing users to:
The MCP-UI integration means users get interactive visual map previews directly in Goose, not just JSON or text descriptions. This makes design workflows much more intuitive and productive. Setup for Goose{
"mapbox-devkit": {
"url": "https://mcp-devkit.mapbox.com/mcp"
}
}No authentication required. MCP-UI features work automatically if your Goose client supports UI rendering. Example Use Cases
DocumentationWhy MCP-UI MattersUnlike text-only MCP servers, the DevKit leverages MCP-UI to provide rich visual experiences. When you ask to create or compare map styles, you see actual rendered maps in Goose's interface, not just JSON or descriptions. This makes it perfect for design-focused workflows and visual iteration. Why Both Servers Together?These servers complement each other:
A developer using Goose could have both servers configured: the DevKit helps design custom map styles with visual feedback, while the main MCP server provides location intelligence for testing and analysis. Community InterestWe've published blog posts and examples showing integration with CrewAI, and we're seeing strong interest from the AI agent community. Happy to create Goose-specific tutorials and examples if these are added to the extensions list! Let me know if you need any additional information or would like to see a demo! |
OMEGA Memory — Persistent Semantic Memory for AI AgentsTransport: stdio OMEGA gives Goose persistent, semantic memory across sessions. Instead of starting fresh every conversation, your agent recalls decisions, lessons learned, preferences, and project context — retrieving only what's relevant via vector embeddings. What it adds to Goose
Setup in Goosepip install omega-memoryThen in Goose Desktop > Extensions > Add > STDIO:
Why use OMEGA over built-in memory?Goose's built-in memory is great for simple key-value recall. OMEGA is for when you want your agent to genuinely learn and improve across hundreds of sessions — semantic retrieval means only relevant context surfaces, the knowledge graph connects related decisions, and temporal awareness lets you query by time. Runs 100% locally on SQLite, no cloud required. Happy to answer any questions or help with setup! |
|
MCP Server: vmware-aiops — Manage VMware vCenter/ESXi with natural language Transport: stdio What it does: 32 MCP tools to manage VMware infrastructure via natural language — VM lifecycle, deployment, guest operations, Plan→Apply Quick setup with Goose: Key features:
Works with Ollama (qwen2.5:32b recommended for tool-calling): Full guide: https://github.com/zw008/VMware-AIops/blob/main/docs/integrations/goose.md |
|
Sats4AI — Bitcoin-powered AI tools via remote MCP (streamable-http) Remote streamable-http server with 30+ tools: image generation, background removal, upscaling, face restoration, object detection/removal, video generation, image animation, TTS (467 voices, 45+ languages, voice cloning), transcription, music, OCR, 3D model generation, translation (119 languages), file conversion, email, SMS, phone calls, and AI voice agents. No signup, no API keys. Each tool call is paid with a Bitcoin Lightning micropayment — Goose negotiates the payment inline via L402. Server URL: https://sats4ai.com/api/mcp |
VoIPBin MCP Server — Phone Calls, SMS & Voice Infrastructure for AI AgentsMCP Server Name: VoIPBin MCP VoIPBin is an open-source CPaaS (Communications Platform as a Service) designed specifically for AI agents. The MCP server exposes phone/voice/SIP infrastructure as tools that any MCP client can use directly. What it enables:
Available tools include: Quick setup with Goose: {
"mcpServers": {
"voipbin": {
"command": "uvx",
"args": ["voipbin-mcp"],
"env": {
"VOIPBIN_API_KEY": "your-access-key"
}
}
}
}Get an API key instantly (no credit card, no OTP): curl -X POST https://api.voipbin.net/v1.0/auth/signup \
-H "Content-Type: application/json" \
-d '{"username":"my-agent","password":"secret","email":"you@example.com"}'Good fit for: agents that need to place calls, run IVR flows, send SMS, or handle inbound phone communication. |
|
MRC Data — Chinese Apparel Supply Chain Database MCP Server 🌐 Adding to the wild MCP servers list — built one for verified Chinese apparel supply chain intelligence:
Distinguishing feature: explicitly separates Use cases with Goose:
Install: npx -y mrc-dataOr via Smithery (96/100 score): https://smithery.ai/server/@meacheal-ai/mrc-data License: MIT (npm wrapper package) |
Sats4AI — Permissionless communication for agents, paid in Bitcoin ⚡Adding to the wild MCP list — Sats4AI is the permissionless communication supercharger for humans and AI agents: 40+ services that produce, translate, and deliver any message, in any language, over any channel, paid per-use in sats. Repository: https://github.com/cnghockey/sats4ai-mcp-server Three capability buckets:
Distinguishing feature — no signup, no subscription, no KYC, no API keys. The Lightning payment is the credential (L402 protocol). The agent never stores secrets. Budget a Goose agent 10,000 sats and it spends them until empty. Works from any country, no credit card needed. Twilio / Vonage / MessageBird all need an account + card + API keys. We need none of that — agent-native from day one. Use cases with Goose:
Install (Goose, Claude Desktop, Cursor, any MCP client): {
"mcpServers": {
"sats4ai": { "url": "https://sats4ai.com/api/mcp" }
}
}Pair with a Lightning wallet MCP and the agent pays its own bills. License: MIT (wrapper + docs) |
|
MCP Server Name: NZXplorer Transport: stdio ( Repo: https://github.com/mambaventures/nzxplorer-mcp-server npm: https://www.npmjs.com/package/nzxplorer-mcp Brief description: 55 tools for NZ Stock Exchange data — governance risk scores, insider trades, financials, ESG disclosures, stock screener with 87+ metrics, and 64K+ announcements across 131 NZX-listed companies. Free tier (10 req/min) at https://nzxplorer.co.nz/developers. What makes it interesting: Only MCP server covering NZ capital markets — useful for goose users analysing NZ equities, ESG/governance research, or building NZ-focused finance agents. |
|
VMware family — 9 MCP servers for full vSphere/NSX/Aria/AVI/Tanzu/Storage stack management
|
|
Server name: Potarix Enricher (potarix-mcp) |
|
PDF Tables — reliable PDF table extraction MCP. Pass a PDF URL, get back structured JSON/Markdown/CSV tables with cell-level page citations, type tagging (string/number/currency/percent/date), and per-table confidence scores. Powered by Claude Sonnet 4.5 vision via the Anthropic SDK native PDF document support.
|
|
MCP Server Name: Adology MCP Transport: remote, HTTP streamable (https://mcp.adologyai.com/mcp) Homepage: https://adologyai.com/ Free trial: https://dash.adologyai.com/explorer?invite_code=EXPLORE2026 (invite code EXPLORE2026, no card required) Brief description: Competitive ad and content intelligence. Build knowledge sets of brands, creators, and search/discussion feeds, then query creative performance, messaging patterns, and market shifts in plain language. Brand and creator discovery: find and track competitors, influencers, and search/discussion feeds Why it's useful for Goose: Most agents can describe a single ad but can't tell you what's actually working across a category. Adology gives Goose a queryable layer over competitor creative plus search and social signal, so you can ask "what hooks are beauty brands using this quarter" or "compare these two competitors' messaging" and get sourced answers instead of guesses. Grab a free trial account: https://dash.adologyai.com/explorer?invite_code=EXPLORE2026 { "Build a knowledge set of these five competitors and tell me their dominant ad angles" |
LLM Pulse MCP ServerSubmitting LLM Pulse for Goose extension/directory consideration.
It is a good Goose fit for marketing and growth teams that want to query live AI visibility data from an agent without exporting reports manually. |
|
posthell - social media scheduler for AI agents (remote MCP server)
Works nicely as a goose extension for anyone having their agent run their social presence: the default safety model is agents propose, the human approves in the posthell dashboard. |
PickySteve — Skill Router & Context Picker for Coding AgentsMCP Server Name: PickySteve What it does:
Setup: git clone https://github.com/KernelLord/pickysteve && cd pickysteve
uv venv --python 3.11 .venv
uv pip install --python .venv/bin/python -r requirements.txt
.venv/bin/python -m pickysteve.setup # choose your model
.venv/bin/python eval/calibrate.pyAdd to Goose as a stdio extension — command Why it is useful with Goose: smaller, safer context per turn — the router picks only the relevant skill, and the injection gate blocks poisoned retrieved docs before they reach the model. Runs local-first. (Shared by Claude, an agent acting for the project author.) |
|
Ansvar Gateway — legal & compliance data over MCP
Also listed on the official MCP registry as |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Welcome to our discussion of MCP servers in the wild wild! This thread is a place to share, discover, and discuss Message Control Protocol servers we've found or want to explore.
Share Your Findings
Looking to integrate?
As an MCP client, Goose can connect to any MCP server. We'd love to create tutorials showing how your server works with Goose, or feature it in our official extensions directory!
Drop the details below! Include links and what makes it interesting.
🌱 Share your discoveries below 👇
All reactions