synthetic-sciences/openscience

TypeScriptopenscience.sh
macOSWindowsLinux

The open-source AI workbench for scientific research

agentaibunclillmmlopen-sourceresearchscienceai-agentml-engineeringresearch-tools
스타 성장
스타
3.5k
포크
468
주간 성장
+138
이슈
9
3k3.5k
26년 7월26년 7월26년 8월26년 9월
아티팩트npm
README
OpenScience

The open-source AI workbench for scientific research

Give it a goal. It reads the literature, writes and runs code, runs the experiments, and writes up what it found.


CI npm release license

Get started · Documentation · Releases · Contributing


OpenScience is an open-source workbench for scientific research. Work through literature, data, code, experiments, and writing in the desktop app, browser workspace, or terminal.

Choose Ace for managed models and research search, connect your own API keys or supported provider sign-in, or use local models. An OpenScience account is optional for your own provider access and local models.

Get started

Download the desktop app for macOS, Windows, or Linux. For the CLI and browser workspace:

npm install -g @synsci/openscience
openscience

Or use the one-step launcher:

npx synsci

Open Customize → Models to choose access. For an existing project:

openscience ~/research/my-project

Follow the quickstart for your first task, or the installation guide for platform details and updates.

Choose model access

Option Setup Usage
Ace Sign in, choose a funding workspace, and fund its Wallet. Pay as you go; no monthly subscription.
Your provider Connect an API key or supported sign-in. Your provider's access and billing terms.
Local model Connect Ollama, LM Studio, or another compatible endpoint. Your hardware; no Ace model charge.

Ace activation is a $0 authorization. While automatic reloads are enabled, a purchased Wallet balance below $5 triggers a $20 reload; payment-processing fees are disclosed separately before payment. Changing to your own provider does not turn off automatic reloads. See Pricing and usage.

Connect from the terminal:

openscience keys add
openscience models

For a local endpoint:

openscience local add

See Models and Local models.

Give it a research task

Inspect data/samples.csv for missing values and inconsistent labels.
Keep the original data unchanged. Save a quality report and a plot
in results/, with the code needed to reproduce them.

Use Research for execution, or start with /plan to agree on the method. Review sources, assumptions, code, and outputs before relying on a scientific conclusion.

For a single terminal turn:

openscience run "Review the analysis plan in this project"
openscience run --continue "Suggest checks for the assumptions you identified"

What you can do

  • Review literature: search scientific sources, compare findings, and save cited evidence.
  • Analyze data: inspect inputs, write and run analysis code, and produce figures and reports.
  • Reproduce experiments: agree on a claim, prerequisites, and budget, then compare measured results.
  • Use scientific capabilities: inspect available tools and connect your own resources where needed.
  • Reuse procedures: browse hundreds of bundled skills or add a project-specific workflow.
  • Extend the workbench: add MCP connections, custom agents and commands, plugins, or an SDK integration.

A skill describes a procedure; it does not mean every referenced tool or service is installed. Check availability in Customize before a substantial task.

Documentation

Browse the capability map, Explore tools, and Skills directory. Each catalog explains what an entry does and links to setup or usage instructions. The workflow cookbook provides complete requests with expected deliverables.

Topic Guide
First use Quickstart, Workspace
Models and billing Models, Ace, Pricing
Research Literature reviews, Data analysis, Reproduction, Writing
Capabilities Skills, Databases, Connectors
Customize Project instructions, Configuration
Automate CLI reference, JSON output, SDK and editors
Help Troubleshooting, FAQ

Plain-text documentation is available at llms.txt and llms-full.txt.

Releases and updates

GitHub Releases contains desktop installers, CLI archives, checksums, and release notes. The changelog records notable behavior changes.

Use the app's update controls or openscience upgrade for the CLI. An npm installation can be updated with npm install -g @synsci/openscience@latest. Back up important project files independently of application updates.

Development

Use the pinned Bun version from package.json:

bun run setup
bun dev
bun run check

CONTRIBUTING.md covers the development loop and required checks. ARCHITECTURE.md maps the codebase, and AGENTS.md describes repository conventions.

For docs changes, see the docs contributor guide and documentation source map.

Control and support

Use the approval controls to decide when OpenScience should ask before acting. Keep original inputs, review consequential actions, and use account controls for spending limits.

Report reproducible product bugs in GitHub Issues. Use account support for private billing questions. Follow SECURITY.md to report a vulnerability.

License

Apache License 2.0. See LICENSE and NOTICE.

OpenScience is an independent project. It is not affiliated with, endorsed by, or sponsored by Anthropic. "Claude" is a trademark of Anthropic, PBC, used here only to describe compatibility.

관련 저장소
Snailclimb/JavaGuide

Java 面试 & 后端通用面试指南,覆盖计算机基础、数据库、分布式、高并发、系统设计与 AI 应用开发

JavaScriptnpminterviewApache License 2.0javainterview
javaguide.cn
158.3k46.1k
langgenius/dify

Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or self-hosted, so teams move from prototype to production without rebuilding the stack.

TypeScriptnpmskillOtheraigpt
dify.ai
154.7k24.4k
TauricResearch/TradingAgents

TradingAgents: Multi-Agents LLM Financial Trading Framework

PythonPyPIApache License 2.0agentfinance
arxiv.org/pdf/2412.20138
102.8k19.8k
OpenHands/OpenHands

🙌 OpenHands: AI-Driven Development

TypeScriptnpmcliMIT Licenseagentartificial-intelligence
openhands.dev
86.4k11.3k
koala73/worldmonitor

Real-time global intelligence dashboard. AI-powered news aggregation, geopolitical monitoring, and infrastructure tracking in a unified situational awareness interface

TypeScriptnpmGNU Affero General Public License v3.0aidashboard
worldmonitor.app
85.7k12.9k
Leonxlnx/taste-skill

Taste-Skill - gives your AI good taste. stops the AI from generating boring, generic slop

JavaScriptnpmskillMIT Licenseagentai
tasteskill.dev
84.9k5.8k
lobehub/lobehub

🤯 LobeHub is your Chief Agent Operator, organizing your agents into 7×24 operations by hiring, scheduling, and reporting on your entire AI team.

TypeScriptnpmskillOtherchatgptopenai
lobehub.com
82.3k15.9k
bytedance/deer-flow

An open-source long-horizon SuperAgent harness that researches, codes, and creates. With the help of sandboxes, memories, tools, skill, subagents and message gateway, it handles different levels of tasks that could take minutes to hours.

PythonPyPIMIT Licenseagentagentic
deerflow.tech
81.7k11.3k
dair-ai/Prompt-Engineering-Guide

🐙 Guides, papers, lessons, notebooks and resources for prompt engineering, context engineering, RAG, and AI Agents.

MDXMIT Licensedeep-learningprompt-engineering
promptingguide.ai
78.1k8.6k
datawhalechina/hello-agents

📚 《从零开始构建智能体》——从零开始的智能体原理与实践教程

PythonPyPItutorialOtheragenttutorial
hello-agents.datawhale.cc
77.3k9.6k
shareAI-lab/learn-claude-code

Bash is all you need - A nano claude code–like 「agent harness」, built from 0 to 1

PythonPyPItutorialMIT Licenseagentclaude-code
learn.shareai.run
76.2k12.3k
unslothai/unsloth

Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4, MiniMax-H3, Gemma 4, FLUX and more.

PythonPyPIApache License 2.0fine-tuningllama
unsloth.ai/docs
75.7k6.9k