Developers · Model Context Protocol

Ask MESSAI from your AI assistant

MESSAI is a Model Context Protocol (MCP) server. Connect it once and your assistant can search 23,596 papers on microbial electrochemical systems, read literature ranges and priors, check where papers disagree and predict reactor performance, citing the papers behind every number.

Server URL

https://www.messai.io/api/mcp

Use the www address. The bare messai.io domain redirects, and some clients do not follow a redirect on a POST request.

Connect in a minute

Claude (claude.ai and the desktop app)

Settings → Connectors → Add custom connector. Name it MESSAI and paste the server URL. Custom connectors are available on Claude plans that support them.

Claude Desktop (config file)

Add this to claude_desktop_config.json, then quit and reopen Claude. It uses the mcp-remote bridge (Node.js required).

{
  "mcpServers": {
    "messai": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://www.messai.io/api/mcp"
      ]
    }
  }
}

Cursor

Use the one-click install, or add this to ~/.cursor/mcp.json:

Add to Cursor

{
  "mcpServers": {
    "messai": {
      "url": "https://www.messai.io/api/mcp"
    }
  }
}

VS Code

Add this to .vscode/mcp.json:

{
  "servers": {
    "messai": {
      "type": "http",
      "url": "https://www.messai.io/api/mcp"
    }
  }
}

Other MCP clients

Add a remote (Streamable HTTP) server with the URL above. Clients that only speak stdio can wrap it with: npx -y mcp-remote <URL>.

Try asking

What is the typical coulombic efficiency of microbial electrolysis cells? Give n and the range.
Benchmark my MFC: carbon-cloth anode, 25 cm², 28 mL, pH 7, 30 °C. Where does it sit in the literature?
Where do papers disagree on internal resistance, and why?
I measured coulombic efficiencies of 0.42, 0.47 and 0.51 in my MEC. How does that change the literature estimate?
Which open research gaps around nitrogen recovery are under-studied?

What it can do

Papers: search the corpus; open a paper by id or DOI with its extracted values and source snippets: search_papers · get_paper
Parameters: find a parameter, its literature distribution per SI unit, the source papers, the Bayesian prior, contradictions, and what it is linked to: list_parameters · get_parameter_distribution · get_parameter_sources · get_prior · get_contradictions · get_parameter_relationships
Modelling: predict reactor performance with calibrated intervals, update a prior with your own measurements, export priors as a dataset: predict_performance · update_with_my_data · export_priors
Discovery: the closest published studies to your conditions; ranked open research gaps: find_similar_studies · find_research_gaps
Reference: electroactive microbes; electrode and membrane materials: lookup_microbes · lookup_materials

Ready-made workflows

Clients that support MCP prompts offer three workflows: benchmark my reactor, review a parameter, and find a research direction.