awesome-repositories.com
Blog
MCP
awesome-repositories.com

Discover the best open-source repositories with AI-powered search.

ExploreCurated searchesOpen-source alternativesSelf-hosted softwareBlogSitemap
ProjectMCP serverAboutHow we rankPress
LegalPrivacyTerms
© 2026 Bringes Technology SRL·VAT RO45896025·hello@awesome-repositories.com
aipoch avatar

aipoch/open-science

0
View on GitHub↗
aipoch.com/open-science
↗

Open Science

Open Science is a local-first AI research workbench in which agents plan and run multi-step scientific tasks on the user's own computer, with results that can be traced back to their code, inputs and environment. Questions, papers or data files given in plain language are worked through by agents that call tools, execute Python and R code in notebook-style cells, and hand tasks to specialist subagents. A separate review context checks each finished turn against replies, execution records and file evidence, and triggers bounded corrections when it finds problems.

What sets it apart is the combination of reproducibility tracking and an open agent platform. Artifacts are stored as immutable versions with their generating code, inputs and environment inventory, can be replayed in isolation to verify outputs, and can be exported as RO-Crate packages. Sessions can also be shared as portable research packages that others replay without running code or calling models. Agents are extended with installable skill packages and MCP tool connectors, and they can run on cloud providers, custom OpenAI-compatible gateways, local model servers, or existing Claude and Codex subscriptions. Python and R code can also be dispatched to remote clusters over SSH or Slurm.

Around that core sits a literature library for importing papers by identifier or link, bringing in Zotero collections, annotating and translating PDFs, running offline OCR, and screening papers against inclusion criteria with AI and human decisions recorded separately. Work is organized into projects and sessions with branches, forks and review-gated plans, and agent permission profiles together with network destination approval control what agents may do. Project data and credentials stay on the local machine.

The same local service backs a desktop app for macOS, Windows and Linux, a browser interface, a headless command-line tool, a dependency-free Node.js SDK, and an MCP server for external AI clients. The command-line tool is distributed through npm, Homebrew, native Linux packages and standalone archives.

Features

  • AI Research Tools - A local-first workbench where AI agents plan, run and review scientific research tasks on the user's own computer with traceable results.
  • Research Agents - Runs multi-step research agents that plan tasks, call tools, and work through questions, papers, or data files on the user's own machine.
  • Scientific Research Platforms - Offers an extensible research agent platform with installable skills, specialist subagents and MCP connectors, usable from desktop, browser, CLI or SDK.
  • Custom Agent Provider Connections - Connects cloud providers, custom compatible gateways, local model servers, or existing Claude and Codex subscriptions and verifies each with a built-in test.

AI search

Explore more awesome repositories

Describe what you need in plain English — the AI ranks thousands of curated open-source projects by relevance.

Start searching with AI
5,490 stars·552 forks·TypeScript·Apache-2.0·3 views
  • Workflow Reproducibility - Traces every result back to its code, inputs and environment, then replays, verifies or shares it as a portable package.
  • Artifact Provenance Management - Stores immutable artifact versions with generating code, inputs, and environment inventory, enabling isolated replay and RO-Crate export for reproducibility.
  • Notebook Execution Environments - Executes Python, R, notebook and shell code in managed or custom environments with background runs, saved history and SSH or Slurm compute.
  • Notebook Code Execution - Runs Python, R, notebook, and shell code locally in managed or custom environments with background execution, saved history, and remote SSH or Slurm dispatch.
  • Artifact Replay Verification - Reruns eligible artifact versions in isolation from their full recipe, inputs and environment, compares outputs, and exports verification records as reproducibility evidence.
  • Local-First Architectures - Keeps project data, credentials, and artifact history on the user's machine, with the desktop app running all logic locally.
  • Agent Backend Selection - Runs work through a choice of agent backends with image input and adjustable reasoning effort in one workspace.
  • Agent Delegation - Lets the main agent hand tasks to installed or user-created specialist agents and subagents, shared through a reviewed marketplace.
  • Agent Review Pipelines - Checks each finished agent turn in a separate review context and triggers bounded corrections with re-review when problems are found.
  • Provider-Agnostic Adapters - Runs the same agent workflows across cloud providers, custom OpenAI-compatible gateways, local model servers and Claude or Codex subscriptions.
  • Skill Installation - Extends agents with built-in and marketplace skills installed or updated in one click, plus skills created from conversations or imported from packages and GitHub.
  • Model Provider Integrations - Connects cloud providers, OpenAI-compatible gateways, local model servers and existing subscriptions so users choose the model and agent backend.
  • Model Context Protocol - Connects external tools and data connectors to agents via the Model Context Protocol and serves research tools outward to third-party clients.
  • MCP Tool Connectors - Links to outside MCP tool servers so agents can call additional data sources and services.
  • Provenance Packaging - Exports an artifact's provenance as an RO-Crate 1.1 package and rebuilds managed environments from complete, compatible environment bundles.
  • Notebook Cell Execution - Executes Python and R code in notebook-style cells so analysis steps run and produce results within the research workflow.
  • Python Data Analysis - Runs Python and R code in notebook-style cells locally or on SSH and Slurm clusters to analyze scientific data with saved history.
  • Research Data Sharing Frameworks - Saves a completed session as a portable package with chosen conversation branches, file versions, notebook records and verification evidence for review, handoff or archiving.
  • Desktop and CLI Distribution Channels - Delivers signed desktop installers with auto-update alongside a localhost web interface, headless CLI, SDK, mobile access, and translated interfaces.
  • Headless CLI and SDK Clients - Provides a headless command-line tool and dependency-free Node.js SDK sharing the same local service, projects, sessions, and permissions as the desktop app.
  • Headless Service Scripting - Drives projects, agent runs and results without a desktop window through a command line, Node.js SDK or external MCP clients.
  • Parallel Branches and Review-Gated Plans - Supports conversation branches, writable forks, side chats, queued messages, and review-gated plans so work can be explored in parallel and approved first.
  • Inclusion Criteria Screeners - Screens papers against inclusion and exclusion criteria with a configured model while keeping AI and manual decisions recorded separately for human review.
  • Literature Collectors - Searches scholarly sources and collects relevant studies for a research question so candidate papers can be screened.
  • Literature Screening and Reference Management - Imports, annotates, translates and screens scholarly papers against inclusion criteria, then organizes references and exports citations.
  • Research Package Replay - Loads a research package into another project or computer as a read-only record replayable step by step without running code, calling models or restoring credentials.
  • Local Data Storage - Keeps project data, settings, artifact versions, provenance and credentials on the local machine, protecting keys with the operating system's secure store.
  • AI Agent Permissions - Limits what agents may edit, run or send over the network through approval, auto-approve-edits or full-access profiles with scoped grants and approved folders.
  • Multi-Kernel Code Execution - Runs Python and R code through notebook-style kernels in managed or user-supplied interpreters, with background execution and remote dispatch over SSH or Slurm.
  • Agent Skill Packages - Defines agent abilities as installable file bundles that can be imported, versioned, and delegated to subagents without changing the core application.
  • Headless Local Services - Runs as a background service reachable through an authenticated browser, CLI, or SDK with no desktop window, sharing projects and permissions.
  • Star history

    Star history chart for aipoch/open-scienceStar history chart for aipoch/open-science

    How this analysis was created: This summary and feature list are AI-generated from collected project material and can contain mistakes. Stars, license and language are imported from GitHub. Inclusion does not mean that we have tested or audited this project. Check the source documentation for any feature you depend on. Learn more on our About page.

    Frequently asked questions

    What does aipoch/open-science do?

    Open Science is a local-first AI research workbench in which agents plan and run multi-step scientific tasks on the user's own computer, with results that can be traced back to their code, inputs and environment. Questions, papers or data files given in plain language are worked through by agents that call tools, execute Python and R code in notebook-style cells, and hand tasks to specialist subagents. A separate review context checks each finished turn against replies,…

    What are the main features of aipoch/open-science?

    The main features of aipoch/open-science are: AI Research Tools, Research Agents, Scientific Research Platforms, Custom Agent Provider Connections, Workflow Reproducibility, Artifact Provenance Management, Notebook Execution Environments, Notebook Code Execution.

    Which projects share features with aipoch/open-science?

    Projects with overlapping indexed features include: microsoft/vscode-copilot-chat — This project is an AI-powered IDE extension and LLM coding assistant that provides a conversational interface for… openai/openai-agents-python — This project is a Python framework for building autonomous, event-driven agent systems. It provides a unified runtime… jtleek/datasharing — This project is a research data sharing framework and provenance protocol designed to ensure computational… i-am-bee/beeai-framework — The BeeAI Framework is an LLM agent framework and multi-agent orchestration engine used to build autonomous agents… huggingface/chat-ui — This project is a web-based user interface for interacting with large language models, featuring streaming responses… alphaxiv/openresearch — OpenResearch is a local-first desktop application that runs coding agents as autonomous research assistants. The…

    Projects sharing features with Open Science

    These projects share indexed features with Open Science. Shared tags can include platform or build tooling; verify the primary use case before treating a result as a replacement.
    • microsoft/vscode-copilot-chatmicrosoft avatar

      microsoft/vscode-copilot-chat

      9,493View on GitHub↗

      This project is an AI-powered IDE extension and LLM coding assistant that provides a conversational interface for generating, refactoring, and debugging code. It functions as an AI agent framework and a Model Context Protocol client, connecting AI models to external data sources and tools to automate complex development tasks. The system is distinguished by its use of autonomous AI agents capable of multi-step task execution, including the ability to read files, modify code, and run terminal commands iteratively. It supports recursive agent orchestration through subagent delegation and employ

      TypeScript
      View on GitHub↗9,493
    • openai/openai-agents-pythonopenai avatar

      openai/openai-agents-python

      27,191View on GitHub↗

      This project is a Python framework for building autonomous, event-driven agent systems. It provides a unified runtime for orchestrating multi-agent workflows, managing persistent conversation state, and executing code within secure, isolated sandbox environments. The framework is designed to handle complex task delegation, allowing agents to invoke other agents as tools while maintaining context across multi-turn interactions. The framework distinguishes itself through its deep integration with the Model Context Protocol, enabling agents to connect to external data sources and remote services

      Pythonagentsaiframework
      View on GitHub↗27,191
    • jtleek/datasharingjtleek avatar

      jtleek/datasharing

      6,796View on GitHub↗

      This project is a research data sharing framework and provenance protocol designed to ensure computational reproducibility. It provides a standardized set of guidelines for transforming raw source data into tidy formats through documented processing scripts and cleaning workflows. The framework distinguishes itself by emphasizing a strict provenance-based packaging system. It requires the organization of raw data, processing recipes, and code books into a single package, ensuring that original unmodified sources are preserved to allow for independent verification of all transformation steps.

      View on GitHub↗6,796
  • huggingface/chat-uihuggingface avatar

    huggingface/chat-ui

    10,766View on GitHub↗

    This project is a web-based user interface for interacting with large language models, featuring streaming responses and persistent conversation history. It functions as an orchestration gateway that directs user prompts to specific language models and acts as a Model Context Protocol client to execute external tools and incorporate live data into conversations. The application includes a routing layer that analyzes input signals and tool requirements to dynamically direct messages to the most appropriate specialized model. It also provides customization settings for brand identity, allowing

    TypeScript
    View on GitHub↗10,766
  • Compare all 30 related projects→