Claude Scientific Writer is a Python-based tool for real-time research lookup, diagram generation, and AI-assisted scientific writing. It supports plugin, CLI, and Python API usage to generate papers, reports, and proposals with integrated data/figures.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
A general purpose scientific writer that combines AI-driven research lookup, diagram generation, and structured document creation. It can generate scientific papers, clinical reports, posters, and grant proposals, with features for literature reviews,BibTeX handling, and figure/data integration. It is usable as a Claude Code plugin, a Python package, or via a native CLI.
How it works
- Real-time research lookup powered by Parallel Search and Extract.
- AI-powered diagram generation from natural-language prompts.
- Intelligent paper detection and bibliographic management.
- Document conversion to Markdown from PDF, DOCX, PPTX, XLSX via MarkItDown.
- Programmatic Python API for asynchronous usage.
Getting started
Prerequisites include Python 3.10+, and API keys for Anthropic and Parallel (via environment variables or login workflow). Installation options:
- Option 2: Install from PyPI (CLI/API Usage)
pip install scientific-writer
# Optional bundled-script runtimes
pip install "scientific-writer[analysis]" # cohort statistics and survival analysis
pip install "scientific-writer[office]" # DOCX/PPTX/XLSX and MarkItDown helpers
- Option 3: Install from source with uv
git clone https://github.com/K-Dense-AI/claude-scientific-writer.git
cd claude-scientific-writer
uv sync
Usage examples cover the plugin, CLI, and Python API, including how to invoke generation and handle outputs. Files placed in data/ are organized into figures/ or data/ automatically. The API demonstrates an async generator yielding updates and results.
Getting started (continued)
- Configure API keys via .env or shell export (ANTHROPIC_API_KEY, PARALLEL_API_KEY, OPENROUTER_API_KEY).
- Example plugin usage demonstrates initialization and command patterns within Claude Code.
- Example CLI usage shows starting a paper and adding data/figures, with outputs saved to writing_outputs/ timestamped topics.
Recent releases
- v2.19.0 (2026-07-29): Refreshed vendored writing-skill subset to upstream Scientific Agent Skills commit ab2f84ab1059 (upstream 2.61.0), regen
- v2.18.0 (2026-07-28): Refreshed vendored writing-skill subset to upstream v2.60.0 at commit 223f5a30f892, regenerated skills.lock.json
- v2.17.1 (2026-07-22): Canonical source pinned to upstream v2.54.0, reproducible synchronization and offline mirror verification
- v2.15.1 (2026-06-15): Model upgrade mentioned for Claude Opus 4.8 as default model across CLI, API, and docs
- v2.14.0 (2026-06-03): Improved citation policies with stricter real-citation requirements
Traction
2169 stars, 257 forks, 4 open issues
Behind the repo
Not provided in the content.
Caveats
MIT license. Created 2025-10-27. Last push 2026-07-29. Language: Python. Typical prerequisites include a LaTeX distribution for PDF generation and optional LibreOffice/FFmpeg for rendering and media conversion.