Scitex Python
Scitex Python: Comprehensive research automation platform — 50+ modules from hypothesis to publication for AI and humans.
Last verified:
What is Scitex Python?
Scitex Python is a comprehensive Python library and research automation platform for AI agents and human researchers. Unified ecosystem of 50+ modules under 'scitex' namespace covering entire research workflow. Key features: @stx.session for reproducible experiments, scitex.io for unified I/O (50+ formats), scitex.plt for publication figures, scitex.scholar for literature search (167M+ papers), and scitex.writer for LaTeX manuscripts. AGPL-3.0 licensed.
Scitex Python pricing
Pricing model: Freemium
SciTeX offers a Freemium model with three subscription tiers. Free tier includes 100 API requests/hour and access to core open-source Python library under AGPL-3.0 license. Premium tier provides 1,000 API requests/hour and includes AI revision assistance, template with color indicators in Writer, and advanced collaboration features. Enterprise tier offers unlimited API requests/hour and advanced collaboration capabilities. The Python library itself is free and open-source with modular installation options (pip install scitex for core only, pip install scitex[all] for everything). SciTeX Cloud can be self-hosted on lab NAS or server at no additional cost.
Scitex Python pros
- Unified toolkit with 50+ modules under single scitex namespace
- 293 MCP tools enabling AI agents to conduct autonomous research
- @stx.session decorator provides auto-CLI, config injection, seed fixation
- Unified file I/O supporting 50+ formats with auto format detection
- Publication-ready figures with reproducible recipes via figrecipe
- 23+ statistical tests with automatic APA-style formatting
- Literature search across 167M+ CrossRef and 250M+ OpenAlex papers
- Automatic PDF download with institutional credentials support
- LaTeX manuscript compilation with BibTeX citation management
- Cryptographic verification via SHA-256 hash-chain DAGs for provenance
- Self-hosted SciTeX Cloud for collaborative research workspace
- Modular installation - install only needed components like plt, stats, scholar
- Auto-exports figure data as CSV alongside plots
- Multi-backend notifications via desktop, phone, SMS, or email
- Open-source AGPL-3.0 license with four freedoms for research
- App SDK for creating and sharing custom domain-specific tools
- Local databases enable offline literature discovery
- Built-in linter for AST-based code pattern checking
- Experimental tracking with frozen parameters and execution logs
Scitex Python cons
- Alpha release (0.17.0-alpha) - data persistence and API stability not guaranteed
- Requires Python 3.10+ which may exclude older systems
- Steep learning curve for researchers new to Python programming
- Complex setup requiring environment configuration with .env.d files
- API rate limits restrict Free tier to only 100 requests/hour
- Some features like CLI and webhooks marked as Coming Soon
- Self-hosting SciTeX Cloud requires NAS or server infrastructure
- Cryptographic verification adds complexity to workflows
- Heavy dependency on external APIs (OpenAI, Anthropic, Google) for AI features
- Local CrossRef/OpenAlex databases require significant storage space
Frequently asked questions about Scitex Python
What is SciTeX and what problem does it solve?
SciTeX is a comprehensive Python library and research automation platform for AI and human researchers. It solves four key problems: (1) fragmented tools - literature search, statistics, figures, and writing each requiring separate incompatible tools; (2) no verification - existing tools address reproducibility but not verification; (3) AI agents lack context - general LLMs cannot operate across the full research lifecycle; (4) no custom tooling - every lab needs domain-specific tools but building them requires deep infrastructure knowledge. SciTeX provides a unified toolkit with 50+ modules, cryptographic verification via Clew, 293 MCP tools for AI agents, and an App SDK for custom tools.
How do I install SciTeX?
SciTeX requires Python 3.10+. Install with: pip install scitex[all] for everything (recommended), pip install scitex for core only (minimal), or install specific modules like pip install scitex[plt,stats,scholar] for typical research setup. Other options include pip install scitex[plt] for figures, pip install scitex[stats] for statistical testing (23+ tests), pip install scitex[scholar] for literature search and PDF download, pip install scitex[writer] for LaTeX compilation, pip install scitex[ai] for LLM APIs, and pip install scitex[cloud] for cloud platform integration. uv is recommended for fast installs.
What is the @stx.session decorator?
The @stx.session decorator provides reproducible experiment tracking with one decorator giving you: auto-CLI generation, YAML config injection, random seed fixation, structured output, and logging. The function docstring becomes --help text. When executed, it creates an output directory with the figure/data files, frozen CONFIGS/CONFIG.yaml with parameters, and logs/{stdout,stderr}.log with execution logs. It also supports notification on completion or failure when notify=True is passed.
How does scitex.scholar work for literature search?
scitex.scholar provides literature discovery and enrichment backed by local CrossRef database (167M+ papers) and OpenAlex database (250M+ works). You can search with stx.scholar.search('topic', n=20) to find papers, fetch papers by DOI with stx.scholar.fetch('10.1038/...'), and enrich BibTeX files with stx.scholar.enrich_bibtex('references.bib', output='enriched.bib'). It supports automatic PDF download with institutional credentials and provides fast full-text search for offline literature discovery.
What statistical tests does scitex.stats support?
scitex.stats supports 23+ statistical tests for publication-ready statistics. You can run tests with stx.stats.run_test('ttest_ind', group1, group2, return_as='dataframe') which returns p-value, effect size (Cohen's d), confidence interval, normality check, and power. It also provides stx.stats.recommend_tests(data) to suggest appropriate statistical tests and stx.stats.format_results(result, style='apa') for APA-style formatting like 't(58) = 2.34, p = .021, d = 0.60'.
How does scitex.clew provide cryptographic verification?
scitex.clew builds a SHA-256 hash-chain DAG linking every manuscript claim back to source data for cryptographic verification in AI-driven science. Every stx.io.load/save automatically records file hashes with zero config. You can check project status with stx.clew.status(), trace files back to source with stx.clew.chain('output.csv'), verify all manuscript claims with stx.clew.dag(claims=True), register traceable assertions with stx.clew.add_claim(), and visualize provenance with stx.clew.mermaid(claims=True). It supports three verification levels: L1 hash comparison (ms), L2 sandbox re-execution (min), and L3 registered timestamp proof (optional).
What file formats does scitex.io support?
scitex.io supports 50+ formats including CSV, JSON, YAML, TOML, HDF5, NPY, NPZ, PKL, PNG, JPG, SVG, PDF, Excel, Parquet, Zarr, INI, TXT, MAT, WAV, MP3, and BibTeX. Format is automatically detected from extension. Built-in features include auto directory creation, path resolution to script_name_out/, symlinks with symlink_from_cwd=True, save logging with file size, and Clew hash tracking. You can also register custom formats with @stx.io.register_saver('.custom') and @stx.io.register_loader('.custom') decorators.
How does scitex.plt create publication-ready figures?
scitex.plt creates reproducible, restylable figures powered by figrecipe. Figures are reproducible nodes in the Clew verification DAG where scientific data and visual style are decomposed, so figures can be restyled (fonts, colors, layout) without altering the underlying data hash. Every figure auto-exports its data as CSV plus a YAML recipe for exact reproduction. Use stx.plt.subplots(), stx.plt.stx_line(), stx.plt.stx_violin(), stx.plt.stx_heatmap(), and stx.io.save(fig, 'analysis.png') which saves analysis.png + analysis.csv + analysis.yaml. You can restyle with stx.plt.reproduce('analysis.yaml', style='nature').
What MCP tools does SciTeX provide for AI agents?
SciTeX provides 293 MCP tools across 23 modules for AI agents to conduct autonomous research. The tool distribution includes: plt (73 tools), cloud (50 tools), writer (38 tools), scholar (22 tools), crossref (15 tools), dev (13 tools), introspect (12 tools), stats (10 tools), clew (9 tools), dataset (8 tools), io (5 tools), notify (5 tools), project (6 tools), tunnel (3 tools), linter (3 tools), social (3 tools), template (4 tools), openalex (4 tools), and docs (4 tools). AI agents can run statistics, create figures, search literature, and compile manuscripts through structured tool calls.
Can I self-host SciTeX Cloud?
Yes, SciTeX Cloud v0.9.3 is a self-hosted research platform that can be deployed on your lab's NAS or server. It provides Scholar, Writer, Console, and Visualizer modules in a unified web interface. The self-hosted version includes a built-in Writer, Scholar, and App Store where researchers build custom tools using scitex-app SDK and scitex-ui components, then share them with the community. A live instance is also hosted at scitex.ai for those who don't want to self-host.