mcp-openidea-search
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@mcp-openidea-searchsearch for papers on transformer architectures"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
mcp-openidea-search
An MCP server that exposes Open Idea's federated search — across open-knowledge providers (OpenAlex, arXiv, Zenodo, Software Heritage, GitHub, GitLab, HuggingFace, PapersWithCode, YouTube, CKAN, Figshare, Kaggle) — to MCP-compatible clients (Claude Desktop, Claude Code, the MCP Inspector, Cursor, etc.).
This package is a thin client over the Open Idea v1 HTTP API. All the federation, deduplication, normalization, and caching happen on Open Idea's side. This repo is fully standalone — no dependency on the Open Idea codebase.
Documentation
Doc | Contents |
Install, configure, and run the server. | |
Components, flows, and a Mermaid diagram. | |
Every tool, resource, and data model. | |
Claude Desktop configuration (macOS/Windows/Linux). | |
Claude Code registration & verification. | |
Testing with the MCP Inspector. | |
401s, newline keys, startup failures, and more. | |
Repo layout, env setup, tests, workflow. |
Related MCP server: Paper Search MCP
🚀 Getting Started
This tool connects Claude directly to Open Idea's academic and data search engines. It runs entirely on your local machine using your personal API key. Follow these 4 simple steps to get set up.
Prerequisites
Make sure you have Python 3.10 or higher installed on your computer. You can
check this by running python --version in your terminal.
Step 1: Get Your Free Open Idea API Key
Because this tool queries live databases on your behalf, you need a personal access key.
Go to openidea.world and sign up or log in.
Navigate to the Developer Dashboard at openidea.world/developer/api-keys.
Click Create New Key.
⚠️ Copy the key immediately (it starts with
oi_live_...). It will only be shown to you once.
Step 2: Install the Search Tool
Open your terminal (Terminal on Mac/Linux, or PowerShell/Command Prompt on Windows) and run the following command to safely install the tool:
pipx install mcp-openidea-search(If you do not have pipx configured, you can use
pip install mcp-openidea-search instead).
Step 3: Connect It to Your Claude Client
Choose the instructions below depending on whether you use Claude Code (in your terminal) or the Claude Desktop App.
Option A: For Claude Code (Terminal)
Close any active Claude sessions by typing
exit.Run this exact command in your terminal (make sure to replace
YOUR_KEY_HEREwith the actual key you copied in Step 1):
claude mcp add openidea \
-e OPENIDEA_API_URL=https://www.openidea.world \
-e OPENIDEA_API_KEY=YOUR_KEY_HERE \
-- mcp-openidea-search stdioOption B: For Claude Desktop (App)
Open your Claude Desktop configuration file. Paste the path below into your file explorer or terminal to open it:
Mac:
~/Library/Application Support/Claude/claude_desktop_config.jsonWindows:
%APPDATA%\Claude\claude_desktop_config.json
Open the file in a text editor (like Notepad or VS Code) and add the following configuration block inside the
"mcpServers"section (replaceYOUR_KEY_HEREwith your actual key):
{
"mcpServers": {
"openidea": {
"command": "mcp-openidea-search",
"args": ["stdio"],
"env": {
"OPENIDEA_API_URL": "https://www.openidea.world",
"OPENIDEA_API_KEY": "YOUR_KEY_HERE"
}
}
}
}Completely restart your Claude Desktop application.
Step 4: Try It Out!
You are all set! You do not need to run this tool manually; Claude will wake it up when needed. Open a chat with Claude and ask a natural question:
💬 "Use the openidea MCP to find recent papers on transformer architectures."
Claude will automatically look up the records, pull data from arXiv, GitHub, or HuggingFace, and summarize the answers for you!
Install
Requires Python 3.10+.
# from PyPI (once published)
pipx install mcp-openidea-search
# or directly from source
pip install -e .Configure
The server reads two environment variables:
Variable | Required | Description |
| yes | Base URL of the Open Idea API ( |
| yes | Personal Bearer key. Mint one at |
Both values are whitespace/newline-trimmed before use.
Run
stdio (Claude Desktop / Claude Code)
Add to ~/Library/Application Support/Claude/claude_desktop_config.json
(macOS) or the equivalent on your OS (see
docs/claude-desktop.md):
{
"mcpServers": {
"openidea": {
"command": "mcp-openidea-search",
"args": ["stdio"],
"env": {
"OPENIDEA_API_URL": "https://www.openidea.world",
"OPENIDEA_API_KEY": "oi_live_xxxxxxxxxxxxxxxxxxxxxxxxxxxx"
}
}
}
}For Claude Code, register it with the CLI (see docs/claude-code.md):
claude mcp add openidea \
-e OPENIDEA_API_URL=https://www.openidea.world \
-e OPENIDEA_API_KEY=oi_live_xxxxxxxxxxxxxxxxxxxxxxxxxxxx \
-- mcp-openidea-search stdioRestart your client and ask: "use the openidea MCP to search for transformer architectures".
HTTP/SSE (remote)
OPENIDEA_API_URL=https://www.openidea.world \
OPENIDEA_API_KEY=oi_live_xxx \
mcp-openidea-search http --host 0.0.0.0 --port 8000Verify with the MCP Inspector:
npx @modelcontextprotocol/inspector http://localhost:8000/sseTools and resources
Tools
Name | Description |
| Search for papers (OpenAlex, arXiv, PapersWithCode). |
| Search for code repos (GitHub, GitLab, Software Heritage). |
| Search for datasets (Zenodo, CKAN, Figshare, Kaggle). |
| Search for ML models (HuggingFace). |
| Search for open hardware. |
| Search for educational videos (YouTube). |
| Search across all providers. Accepts an optional |
Each tool takes query: string and optional limit: integer (default 20,
range 1–100). Returns a JSON payload { results: Resource[], total: number, … }.
Full schemas and response shapes are in
docs/api-reference.md.
Resources
Single items addressable by URI (openidea://<provider>:<id>):
openidea://openalex:W2741809807
openidea://github:vercel/next.jsListing resources returns an empty array — they're addressable but not
enumerable. Use the search tools to discover ids, then read_resource to fetch
the canonical record.
Develop
python -m venv .venv
source .venv/bin/activate # Windows: .venv\Scripts\Activate.ps1
pip install -e ".[dev]"
pytest -qSee docs/development.md for the full contributor guide.
Troubleshooting
Symptom | Likely cause |
| Key is missing, malformed, has a stray newline, or has been revoked. Mint a new one. |
| Throughput limit on this key (server retries once). Wait or use a second key. |
Empty results | Upstream provider down, or |
|
|
Full guide: docs/troubleshooting.md.
License
MIT.
This server cannot be installed
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Latest Blog Posts
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/Sony17/OpenIdeaMcp-'
If you have feedback or need assistance with the MCP directory API, please join our Discord server