Analysis updated 2026-05-18
Chat with an AI agent that can read, summarize, and compare research papers
Browse and preview scientific files like PDFs, molecular structures, and datasets in one workspace
Enable reusable skills for literature search, figures, or domain specific research workflows
Approve an agent's request to run a compute job on a remote Linux server over SSH
| qzzqzzb/openclaudescience | 2latemc/justanothermusicclient | io-software-ai/palserver-gui | |
|---|---|---|---|
| Stars | 82 | 82 | 82 |
| Language | TypeScript | TypeScript | TypeScript |
| Setup difficulty | moderate | moderate | easy |
| Complexity | 4/5 | 3/5 | 3/5 |
| Audience | researcher | general | general |
Figures from each repo's GitHub metadata at analysis time.
Requires configuring a model provider API key and, for remote compute, an SSH host from your local SSH config.
InternAgentS is an open source attempt to recreate the experience of Anthropic's Claude Science tool for researchers, but built so it can run on your own machine instead of relying on a fixed cloud service. It is a chat based workspace where an AI agent helps with real scientific work: reading papers, analyzing results, making figures, writing documents and slides, and running research specific workflows, all built on top of two existing agent frameworks called DeepAgents and LangGraph. The interface is organized into three parts: a left sidebar for navigating projects, sessions, and settings, a center panel for chatting with the agent, and a right panel showing your project's files, previews, and information about what the agent is doing behind the scenes. Instead of the agent grabbing files directly off your computer, it goes through a defined workspace system, so file browsing, search, and previews of things like PDFs, Office documents, images, and even molecular structures all work in a consistent, controlled way. A key idea is reusable skills, meaning packaged capabilities such as literature search or result analysis that can be turned on for a project or a single conversation. It also supports connecting outside tools through a plugin style protocol, and lets you choose which AI model powers the agent, whether that is a cloud service, a private gateway, or a model running on your own hardware. One more advanced feature lets you register a remote Linux computer over SSH, so the agent can send it larger compute jobs to run in the background. Nothing runs automatically here: the agent proposes the job, it shows up in the chat as something you have to explicitly approve, and only then does it get submitted and its results collected back into your workspace. The stated priority throughout is keeping your files, credentials, and workspace state on machines you control by default, rather than requiring any particular company's cloud infrastructure. It uses Python and TypeScript with a Next.js interface, and is released under the MIT license.
A local first, open source research workspace where an AI agent helps you read papers, analyze data, and manage scientific files, similar to Claude Science but self hostable.
Mainly TypeScript. The stack also includes Python, TypeScript, Next.js.
Use freely for any purpose, including commercial use, as long as you keep the copyright notice.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly researcher.
This repo across BitVibe Labs
Verify against the repo before relying on details.