ScienceClaw
ScienceClaw is a personal research assistant built with LangChain DeepAgents and AIO Sandbox infrastructure, adopting a completely new architecture beyond OpenC
What is ScienceClaw?
ScienceClaw is a personal research assistant built with LangChain DeepAgents and AIO Sandbox infrastructure, adopting a completely new architecture beyond OpenClaw. It offers stronger security, better transparency, and a more user-friendly experience.
Overview of ScienceClaw
What it is
ScienceClaw is a personal research assistant built with LangChain DeepAgents and AIO Sandbox infrastructure. It adopts a completely new architecture beyond OpenClaw. It offers stronger security, better transparency, and a more user-friendly experience. It includes 1,900+ built-in scientific tools, supports multi-format content generation, and is fully local and privacy-first.
Key features
- Runs entirely inside Docker containers in an isolated sandbox; agent cannot access host system, personal files, or environment variables.
- Every step of the agent's workflow is visible and traceable, from web search to final report generation.
- Ships with curated tools and skill packages; launch the entire environment with a single command.
- 1,900+ built-in scientific tools spanning drug discovery, astronomy, earth science, chemistry, biodiversity, social science, academic literature, and data science.
- Four-layer tool architecture: built-in tools, ToolUniverse, sandbox tools, and custom @tool functions.
- Skill system with structured instruction documents (SKILL.md) for complex multi-step workflows; built-in skills for PDF, DOCX, PPTX, XLSX, tool creation, skill creation, skill discovery, and unified tool access.
- Multi-format report generation: PDF, DOCX, PPTX, XLSX.
Use cases
ScienceClaw is designed for researchers, developers, and students who need a secure, transparent, and easy-to-deploy research assistant. It can be used for scientific literature review, data analysis, report generation, and multi-step research workflows across various scientific domains.
Getting started
Windows users: Download the desktop installer (ScienceClaw Desktop v0.0.4), extract and run the installer, then launch from desktop shortcut.macOS / Linux users:Prerequisites: Docker & Docker Compose, system RAM ≥ 8 GB.bashgit clone https://github.com/AgentTeam-TaichuAI/ScienceClaw.gitcd ScienceClawdocker compose -f docker-compose-release.yml up -d --pull alwaysOpen http://localhost:5173 in browser. Default login: username admin, password admin123. Change password after first login.Developers: Build from source with docker compose up -d --build.
License and pricing
The license is not stated. The project is open source (pricing_type: open_source). Free LLM API credits are available for early users: SCNet offers 10M free tokens, Zidong Taichu Cloud offers 10M free tokens, limited availability.
Based on the project's public README and repository metadata.