Lacuna Research MCP — Claude Code MCP server: what it does and setup
Links coding agents to a machine learning knowledge graph. It fetches papers, research proposals, author dossiers, and academic venues directly in your editor.
claude mcp add --transport http lacuna-research-mcp https://lacuna.tiptreesystems.com/api/mcpWhat Lacuna Research MCP does
This server connects your development environment to Lacuna's scholarly database. The index integrates records from OpenAlex, OpenReview, DBLP, and arXiv. It tracks over 2.29 million papers, 215,000+ author profiles, and 25,000+ generated research proposals. Every indexed record traces back to an exact document and source page.
Agents query concept clusters to inspect emerging directions or pull literature formatted in clean Markdown. The tool also maps academic institutions and conference publication patterns. This helps teams inspect theoretical backgrounds without opening an external browser.
The service runs as a hosted Streamable HTTP endpoint at https://lacuna.tiptreesystems.com/api/mcp without local dependencies. Running locally via uvx demands Python 3.11 or later. Early prerelease builds of Python 3.14 will fail due to MCP SDK compatibility limits.
What it helps with
- retrieve recent ML papers in markdown with source page citations
- investigate publication records and collaborator networks for an author
- explore generated research proposals across machine learning clusters
- evaluate activity trends across AI conference venues and institutions
Example prompts for the agent
- Search arXiv literature on sparse attention mechanisms and include page references
- Find the main research directions and collaborator graph for author Yoshua Bengio
- Inspect generated research proposals linked to the diffusion model concept cluster
How to install Lacuna Research MCP
1Platform CLI: remote server
claude mcp add --transport http lacuna-research-mcp https://lacuna.tiptreesystems.com/api/mcp2Platform CLI
claude mcp add lacuna-research-mcp -- uvx lacuna-research-mcpuvx runs the command named after the package; if the README names another one, use uvx --from lacuna-research-mcp <command>
1Platform CLI: remote server
codex mcp add lacuna-research-mcp --url https://lacuna.tiptreesystems.com/api/mcp2Platform CLI
codex mcp add lacuna-research-mcp -- uvx lacuna-research-mcpuvx runs the command named after the package; if the README names another one, use uvx --from lacuna-research-mcp <command>
Config file
{
"mcpServers": {
"lacuna-research-mcp": {
"url": "https://lacuna.tiptreesystems.com/api/mcp"
}
}
}{
"mcpServers": {
"lacuna-research-mcp": {
"command": "uvx",
"args": [
"lacuna-research-mcp"
]
}
}
}uvx runs the command named after the package; if the README names another one, use uvx --from lacuna-research-mcp <command>
Config file
{
"servers": {
"lacuna-research-mcp": {
"type": "http",
"url": "https://lacuna.tiptreesystems.com/api/mcp"
}
}
}{
"servers": {
"lacuna-research-mcp": {
"type": "stdio",
"command": "uvx",
"args": [
"lacuna-research-mcp"
]
}
}
}uvx runs the command named after the package; if the README names another one, use uvx --from lacuna-research-mcp <command>
1Platform CLI: remote server
gemini mcp add -t http lacuna-research-mcp https://lacuna.tiptreesystems.com/api/mcp2Platform CLI
gemini mcp add lacuna-research-mcp uvx lacuna-research-mcpuvx runs the command named after the package; if the README names another one, use uvx --from lacuna-research-mcp <command>
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Lacuna Research MCP
How do you install Lacuna Research MCP in Claude Code?
Run the remote connector command in your terminal: claude mcp add --transport http lacuna-research-mcp https://lacuna.tiptreesystems.com/api/mcp.
What does Lacuna Research MCP do?
It grants agents access to a structured research graph linking papers, hypotheses, authors, and venues. The agent can summarize academic literature, trace citations, and explore research directions during coding sessions.
What are the requirements for running the server locally?
You need the uv tool and Python 3.11 or newer. Early Python 3.14 prereleases are unsupported by the MCP SDK, so use an official release like Python 3.13.
