research-hub — Claude Code MCP server: features and setup
Connect literature, research evidence, results, and manuscripts in a local workspace. Hand tasks to AI and review its proposals.
claude mcp add research-hub -- uvx research-hub-pipelineWhat research-hub does
research-hub organizes research projects. Track references and inclusion decisions, link claims to sources, record methods and results, and prepare manuscript outlines or revisions. You can register Word, LaTeX, and Markdown documents. The workspace keeps files local and records their paths and hashes; your editor remains in charge of changing the originals.
Add the server to Claude Code with claude mcp add research-hub -- uvx research-hub-pipeline. The workspace preview needs Python 3.10+ and Git; the published package can also run a sample dashboard. Tasks can move between the workspace and an AI host through export and import. Connected Codex can draft prose for human review, using text excerpts rather than binary document contents. Handle Word formatting and LaTeX compilation in your editor.
Changed registered files can make a task stale, and proposals wait for review. Acceptance is a separate decision; the workflow stages do not advance automatically.
What it helps with
- connect a research question to references and supporting evidence
- record why a publication was included or excluded
- prepare a manuscript outline and hand a task to an AI host
- review a proposal and record whether to accept or revise it
Example prompts for the agent
- Set up a local research project with its question, references, and inclusion decisions.
- Link the claims in this project to their sources and show which support still needs review.
- Draft a manuscript proposal for human review and identify the materials it draws on.
How to install research-hub
Platform CLI
claude mcp add research-hub -- uvx research-hub-pipelineuvx runs the command named after the package; if the README names another one, use uvx --from research-hub-pipeline <command>
Platform CLI
codex mcp add research-hub -- uvx research-hub-pipelineuvx runs the command named after the package; if the README names another one, use uvx --from research-hub-pipeline <command>
Config file
{
"mcpServers": {
"research-hub": {
"command": "uvx",
"args": [
"research-hub-pipeline"
]
}
}
}uvx runs the command named after the package; if the README names another one, use uvx --from research-hub-pipeline <command>
Config file
{
"servers": {
"research-hub": {
"type": "stdio",
"command": "uvx",
"args": [
"research-hub-pipeline"
]
}
}
}uvx runs the command named after the package; if the README names another one, use uvx --from research-hub-pipeline <command>
Platform CLI
gemini mcp add research-hub uvx research-hub-pipelineuvx runs the command named after the package; if the README names another one, use uvx --from research-hub-pipeline <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 research-hub
How do I install research-hub in Claude Code?
Run claude mcp add research-hub -- uvx research-hub-pipeline to add the research-hub MCP server.
What does research-hub do?
It organizes references, evidence, results, and manuscripts in a local research workspace. People review task outcomes and AI proposals.
What do I need to preview the workspace?
The preview requires Python 3.10+ and Git. Its commands create a virtual environment and start the workspace server with the workspace-demo directory.
