Rigorously

Automated research quality assurance.

Rigorously catches the mistakes that slip past manual review — fabricated citations, overclaimed results, irreproducible numbers, and statistical misinterpretations. One command. Eight checks.

Tested on: Python CLI, Claude Code, pre-commit hooks · Compatible with: 16+ AI coding platforms via the Agent Skills standard and MCP

The Problem

Citation errors appear in 25% of published papers. "Statistically significant" gets misused in half of biomedical literature. Overclaimed results are the #1 reason reviewers reject computational papers. Manual review catches some of these. Rigorously catches the rest.

pip install rigorously
rigorously check paper.tex

What It Catches

Sample Output

$ rigorously check paper.tex

  Overclaim Detection: 4 findings
  ┌──────────┬────────┬───────────────────────┬─────────────────────────────┐
  │ CRITICAL │ L.56   │ validated             │ Use "consistent with" unless│
  │          │        │                       │ quantitatively compared     │
  │ CRITICAL │ L.216  │ proof_language         │ Models provide evidence,    │
  │          │        │                       │ not proof                   │
  │ WARNING  │ L.502  │ confirms_demonstrates │ Models predict or suggest;  │
  │          │        │                       │ they do not confirm         │
  │ INFO     │ L.89   │ significant_ambiguous  │ Specify p < X or use       │
  │          │        │                       │ "substantial"               │
  └──────────┴────────┴───────────────────────┴─────────────────────────────┘

  Citation Verification: 12 entries checked
    ✓ Best2010 — DOI resolves, metadata matches
    ✓ Karin2020 — DOI resolves, metadata matches
    ✗ LePoul2000 — Author mismatch: bib has "Bhatt" x4, PubMed has "Hanoun"

  VERDICT: 4 critical issues. Fix before submission.

Platform Support

How It Works

paper.tex + refs.bib
       │
       ▼
  ┌─────────┐    ┌──────────┐    ┌──────────┐
  │  Parse   │───▶│ Extract  │───▶│ Verify   │
  │ LaTeX/MD │    │ claims,  │    │ against  │
  │          │    │ numbers, │    │ CrossRef,│
  │          │    │ citations│    │ code,    │
  │          │    │          │    │ PubMed   │
  └─────────┘    └──────────┘    └──────────┘
                                       │
                                       ▼
                                ┌──────────┐
                                │  Report  │
                                │ GO/NO-GO │
                                └──────────┘

MCP Server

For AI agent integration:

pip install "rigorously[mcp]"
python -m rigorous.mcp_server

Tools: check_paper, verify_citation, check_overclaims, audit_parameters, generate_report

Pre-commit Hook

rigorously install-hook
# Blocks commits with critical integrity issues in paper files

vs. Alternatives

Origin

Built to catch real mistakes in real research. During development of a computational neuroscience paper, Rigorously caught 5 fabricated bibliography entries, 4 overclaimed results, and a parameter bug that was disguised as a scientific discovery — all before submission. It now runs on every commit.

Contributing

See CONTRIBUTING.md.

Citation

If you use Rigorously in your research workflow:

@software{rigorously,
  author = {Miraj, Mansib},
  title = {Rigorously: Automated Research Quality Assurance},
  url = {https://github.com/XenResearch/rigorously},
  year = {2026}
}

License

MIT