All ClawAll Claw
Back to Catalog
sciclaw logo
client

sciclaw

sciclaw is a lightweight, open-source paired-scientist AI assistant that turns your local filesystem into a reproducible research co-worker for literature review, hypothesis-driven experimentation, manuscript drafting, and more.

Overview

sciclaw (often styled as sciClaw) is an open-source, paired-scientist AI assistant designed specifically for rigorous scientific research workflows. Built as a lightweight Go binary (~17 MB), it extends frameworks like PicoClaw to act as an autonomous research co-worker that lives inside your project directory.

It connects to major LLM providers (or bring-your-own models), proposes and executes hypothesis-driven research loops, handles literature searches (e.g., PubMed), drafts and revises manuscripts in Word with tracked changes, manages references, and logs every action for full reproducibility.

Key Features

  • Lightweight Local Runtime: Runs as a native Go binary on macOS, Linux, and Windows (including WSL) with minimal dependencies.
  • Scientific Skills: Includes 12+ baseline research-focused tools such as PubMed search, experiment provenance tracking, manuscript co-authoring, and data analysis support.
  • Reproducibility First: Automatically logs all actions, decisions, and outputs in a structured workspace for verifiable science.
  • Manuscript Integration: Directly edits Microsoft Word documents with tracked changes and handles reference management.
  • LLM Flexibility: Supports ChatGPT, Claude, and other providers; no vendor lock-in.
  • Lifecycle Hooks & Autonomy: Hypothesis proposal, experiment design, execution, and iterative improvement in closed-loop workflows.
  • Easy Installation: Available via pre-compiled binaries, source build, or Docker.

Installation

Binary Download

Pre-compiled binaries for multiple platforms are available on the GitHub Releases page.

Homebrew / Source

git clone https://github.com/drpedapati/sciclaw.git
cd sciclaw
make deps && make install

Docker Support

Full Docker setup instructions are provided in the repository for sandboxed execution.

Use Cases

  • Literature Review & Synthesis: Automated PubMed searches and summary generation.
  • Experiment Planning & Execution: Hypothesis-driven loops with provenance tracking.
  • Manuscript Preparation: AI-assisted drafting and revision in Word with change tracking.
  • Reproducible Workflows: Full audit logs for academic publishing and collaboration.
  • Daily Research Companion: Acts as a persistent co-worker embedded in your project filesystem.

Technical Details

sciclaw is MIT-licensed and emphasizes privacy and control by running locally. It integrates deeply with your filesystem, turning directories into intelligent research environments while maintaining strict logging for scientific integrity.

A setup tutorial video is available on YouTube for quick onboarding.

Related Projects

sciclaw builds upon or relates to the broader OpenClaw/PicoClaw ecosystem of local AI agents.

Tags

airesearchscientificagentgoopen-sourcereproducible-researchpubmedmanuscriptllm