Scitex

How to Set Up and Use Scitex

SciTeX is a comprehensive Python toolkit for reproducible scientific research that covers the entire pipeline from raw data ingestion to manuscript submission. It exposes 323 MCP tools across 23 modules — including publication-ready figure generation (`plt`), statistical testing, literature search and PDF enrichment (`scholar`), LaTeX manuscript compilation (`writer`), and cryptographic reproducibility verification (`clew`) — so AI assistants like Claude can autonomously conduct analyses, generate figures, search papers, and verify that manuscript claims trace back to source data. Researchers use it to eliminate manual figure editing, automate citation management, and produce verifiably reproducible papers.

uv pip install 'scitex[all]'
# Lighter alternatives:
# uv pip install 'scitex[plt,stats,scholar]'  # Typical research
# uv pip install 'scitex[writer]'              # LaTeX only

cp -r .env.d.examples .env.d
# Edit credentials:
$EDITOR .env.d/
source .env.d/entry.src

scitex mcp list-tools

{
  "mcpServers": {
    "scitex": {
      "command": "scitex",
      "args": ["mcp", "start"],
      "env": {
        "SCITEX_ENV_SRC": "/path/to/your/.env.d/entry.src"
      }
    }
  }
}

- "Generate a publication-ready bar chart from results.csv with error bars and export it as both PNG and PDF."
- "Search for recent papers on neural oscillations and download the top 5 as PDFs with enriched metadata."
- "Run appropriate statistical tests on my dataset and recommend which test to use based on the data distribution."
- "Compile my LaTeX manuscript and check that all figure claims are cryptographically verified against source data."
- "Show me the dependency graph for figure_3.png and trace it back to its source data file."