[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-73832":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":9,"language":10,"languages":9,"totalLinesOfCode":9,"stars":11,"forks":12,"watchers":13,"openIssues":13,"contributorsCount":14,"subscribersCount":14,"size":14,"stars1d":15,"stars7d":16,"stars30d":17,"stars90d":14,"forks30d":14,"starsTrendScore":18,"compositeScore":19,"rankGlobal":9,"rankLanguage":9,"license":20,"archived":21,"fork":21,"defaultBranch":22,"hasWiki":21,"hasPages":21,"topics":23,"createdAt":9,"pushedAt":9,"updatedAt":24,"readmeContent":25,"aiSummary":26,"trendingCount":14,"starSnapshotCount":14,"syncStatus":27,"lastSyncTime":28,"discoverSource":29},73832,"DeepScientist","ResearAI\u002FDeepScientist","ResearAI","Now, Stronger AI Pushes Frontiers, Stronger Our Shared Future.",null,"TypeScript",3014,318,14,0,78,157,423,234,109.51,"Apache License 2.0",false,"main",[],"2026-06-12 04:01:12","\u003Ch1 align=\"center\" style=\"font-size: 3.25rem; line-height: 1.02; margin-bottom: 0.4rem;\">\n  \u003Cimg src=\"assets\u002Fbranding\u002Flogo.svg\" alt=\"DeepScientist logo\" width=\"50\" \u002F>\n  DeepScientist\n\u003C\u002Fh1>\n\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002FResearAI\u002FDeepScientist\">GitHub\u003C\u002Fa> |\n  \u003Ca href=\"README_ZH.md\">中文文档\u003C\u002Fa> |\n  \u003Ca href=\"docs\u002Fen\u002FREADME.md\">English Docs\u003C\u002Fa> |\n  \u003Ca href=\"https:\u002F\u002Fopenreview.net\u002Fforum?id=cZFgsLq8Gs\">Paper\u003C\u002Fa> |\n  \u003Ca href=\"https:\u002F\u002Fdeepscientist.cc\u002F\">Website\u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002FResearAI\u002FDeepScientist\">\u003Cimg alt=\"GitHub stars\" src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002FResearAI\u002FDeepScientist?style=for-the-badge&logo=github\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fevent.baai.ac.cn\u002Factivities\u002F962\">\u003Cimg alt=\"Watch Video\" src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FWatch-Video-5B7266?style=for-the-badge\">\u003C\u002Fa>\n  \u003Ca href=\"LICENSE\">\u003Cimg alt=\"License Apache-2.0\" src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-Apache%202.0-yellow.svg?style=for-the-badge\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fwww.python.org\u002F\">\u003Cimg alt=\"Python 3.11+\" src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FPython-3.11%2B-blue?style=for-the-badge&logo=python&logoColor=white\">\u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fopenreview.net\u002Fforum?id=cZFgsLq8Gs\">\u003Cimg alt=\"ICLR 2026 Top 10 Badge\" src=\"assets\u002Freadme\u002Ficlr2026_top10_badge.svg\" height=\"44\">\u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Cstrong>15-minute local setup\u003C\u002Fstrong> ·\n  \u003Cstrong>One repo per quest\u003C\u002Fstrong> ·\n  \u003Cstrong>Visible research progress\u003C\u002Fstrong> ·\n  \u003Cstrong>Human takeover anytime\u003C\u002Fstrong>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Cstrong>Built-in runners: Codex, Claude Code, Kimi Code, OpenCode\u003C\u002Fstrong>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"docs\u002Fen\u002F00_QUICK_START.md\">Quick Start\u003C\u002Fa> •\n  \u003Ca href=\"docs\u002Fen\u002F02_START_RESEARCH_GUIDE.md\">Launch Your First Project\u003C\u002Fa> •\n  \u003Ca href=\"docs\u002Fen\u002F12_GUIDED_WORKFLOW_TOUR.md\">Product Tour\u003C\u002Fa> •\n  \u003Ca href=\"docs\u002Fen\u002F15_CODEX_PROVIDER_SETUP.md\">Codex Setup\u003C\u002Fa> •\n  \u003Ca href=\"docs\u002Fen\u002F24_CLAUDE_CODE_PROVIDER_SETUP.md\">Claude Setup\u003C\u002Fa> •\n  \u003Ca href=\"docs\u002Fen\u002F27_KIMI_CODE_PROVIDER_SETUP.md\">Kimi Setup\u003C\u002Fa> •\n  \u003Ca href=\"docs\u002Fen\u002F25_OPENCODE_PROVIDER_SETUP.md\">OpenCode Setup\u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  Maintainers: \u003Ca href=\"docs\u002Fen\u002F22_BENCHSTORE_YAML_REFERENCE.md\">BenchStore YAML Guide\u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Cstrong>May 12 update:\u003C\u002Fstrong> v1.6.0 is available with Claude Code, OpenCode, Kimi Code, BenchStore, and science evidence workflows.\n\u003C\u002Fp>\n\n![deepscientist_install](https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002Fd8244944-4f70-4e08-94e3-002b74ce70fb)\n\nUnlike one-shot **AI Scientist** or **autoresearch-style systems**, DeepScientist is a **local-first autonomous research studio** that keeps the full loop moving on your machine, from **baselines** and **experiment rounds** to **paper-ready outputs**, with a **10-minute setup**. Powered by **Findings Memory**, **Bayesian optimization**, and the **Research Map**, it keeps turning each new result into the next starting point and goes deep through broader exploration and, when needed, **thousands of experiment validations**.\n\nIf you want the technical deep dive behind DeepScientist, watch the [Video](https:\u002F\u002Fevent.baai.ac.cn\u002Factivities\u002F962).\n\n---\n\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F3c7abb44-2b25-4477-a011-10a3154d6d76\n\n## Still Spending Your Time On Research Grunt Work?\n\nWhat drains researchers is often not the lack of ideas. It is the endless cycle of low-leverage work:\n\n- new papers keep coming, but only a small fraction turns into an actionable next-step research plan\n- baseline repos fail on environment, dependency, data, and script issues before real work even starts\n- experiment results get scattered across terminals, scripts, notes, and chats, making later review painful\n- writing, figures, and analysis live in separate tools, so turning them into a coherent paper takes far too long\n\nThis is the problem DeepScientist is built to solve:\n\n> turn fragmented, repetitive, easy-to-lose research work into a local AI workspace that can keep moving, keep accumulating, and keep getting stronger over time\n\n## DeepScientist Is Not Just Another \"Research Chatbot\"\n\nIt is not a tool that summarizes papers, throws you a few ideas, and leaves the dirty work to you.\n\nIt is much closer to a real long-running AI research partner:\n\n| What common AI tools often look like | What DeepScientist does instead |\n|---|---|\n| Great at chatting, but context disappears quickly | Turns tasks, files, branches, artifacts, and memory into durable state |\n| Good at suggesting ideas, but weak at sustained execution | Pushes papers, baselines, experiments, and writing inside one workspace |\n| Strong automation, but feels like a black box | Lets you inspect the process through the web workspace, Canvas, files, and terminal |\n| Hard to take over once it goes off track | Lets you pause, take over, edit plans, change code, and continue at any time |\n| Each run ends when the run ends | Preserves failed paths, winning paths, and reproduction lessons for the next round |\n\n## About\n\n> DeepScientist is not a one-shot agent demo. It is a system built for long-horizon research work.\n\n## What Can It Actually Help You Get Done?\n\n### 1. Start a real project from a paper or a research question\n\n- feed it a core paper, a GitHub repository, or a natural-language research objective\n- it turns those inputs into an executable quest instead of a chat that loses state after a few turns\n\n### 2. Reproduce baselines and keep the reproduction reusable\n\n- restore repositories, prepare environments, handle dependencies, and track the critical failures\n- preserve what broke, what got fixed, and which steps are trustworthy for future rounds\n\n### 3. Run experiments continuously instead of stopping after one pass\n\n- propose the next hypothesis from existing results\n- branch, ablate, compare, and record conclusions\n- keep failed routes as assets instead of deleting them\n\n### 4. Turn results into materials you can actually ship\n\n- organize findings, conclusions, and analysis\n- produce figures, reports, and paper drafts\n- support local PDF and LaTeX compilation workflows\n\n### 5. Follow the same research effort from multiple surfaces\n\n- the web workspace in your browser\n- the TUI workflow on a remote server\n- external connector surfaces for collaboration and progress updates\n\nThe current docs already cover these collaboration channels:\n\n- [Weixin](docs\u002Fen\u002F10_WEIXIN_CONNECTOR_GUIDE.md)\n- [QQ](docs\u002Fen\u002F03_QQ_CONNECTOR_GUIDE.md)\n- [Telegram](docs\u002Fen\u002F16_TELEGRAM_CONNECTOR_GUIDE.md)\n- [WhatsApp](docs\u002Fen\u002F17_WHATSAPP_CONNECTOR_GUIDE.md)\n- [Feishu](docs\u002Fen\u002F18_FEISHU_CONNECTOR_GUIDE.md)\n- [Lingzhu \u002F Rokid](docs\u002Fen\u002F04_LINGZHU_CONNECTOR_GUIDE.md)\n\n## Why Is It Easier To Keep Using?\n\nWhat retains users is not a flashy demo. It is a system that becomes more useful the longer you work with it.\n\nDeepScientist tends to stick for four reasons:\n\n### Local-first by default\n\n- code, experiments, drafts, and project state stay on your own machine or server by default\n- this is especially valuable for unpublished ideas, sensitive experiment history, and longer-running research loops\n\n### One repo per quest\n\n- every quest is a real Git repository\n- branches, worktrees, files, and artifacts naturally express research structure\n\n### The process is not a black box\n\n- it does not only give you an output\n- you can inspect what it read, what it changed, what it kept, and what it plans to do next\n\n### Human collaboration is built in\n\n- DeepScientist can move autonomously\n- you can also step in, edit, redirect, and hand control back whenever you want\n\n## Why Try It Now?\n\nBecause this is not just a concept. It is a real system with public docs, a public paper, and a public install path.\n\n- `2026\u002F03\u002F24`: DeepScientist officially released `v1.5`\n- `2026\u002F02\u002F01`: the paper went live on [OpenReview](https:\u002F\u002Fopenreview.net\u002Fforum?id=cZFgsLq8Gs) for `ICLR 2026`\n- npm install path is already available: [`@researai\u002Fdeepscientist`](https:\u002F\u002Fwww.npmjs.com\u002Fpackage\u002F@researai\u002Fdeepscientist)\n- both Chinese and English docs are available, along with Web, TUI, and connector entry points\n\n## Product Preview\n\n### Architecture Overview\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"assets\u002Freadme\u002Farchitecture-promo.png\" alt=\"DeepScientist architecture overview\" width=\"92%\" \u002F>\n\u003C\u002Fp>\n\n### Example Outputs\n\n\u003Ctable>\n\u003Ctr>\n\u003Ctd width=\"50%\">\n\u003Cimg src=\"assets\u002Freadme\u002Fpaper-output-1.png\" alt=\"DeepScientist generated paper example 1\" width=\"100%\" \u002F>\n\u003C\u002Ftd>\n\u003Ctd width=\"50%\">\n\u003Cimg src=\"assets\u002Freadme\u002Fpaper-output-2.png\" alt=\"DeepScientist generated paper example 2\" width=\"100%\" \u002F>\n\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd valign=\"top\">\n\u003Cb>Example paper output 1\u003C\u002Fb>\u003Cbr\u002F>\nPaper-facing deliverables can be preserved directly inside the quest instead of being split across external tools.\n\u003C\u002Ftd>\n\u003Ctd valign=\"top\">\n\u003Cb>Example paper output 2\u003C\u002Fb>\u003Cbr\u002F>\nDeepScientist can carry work through writing, review, figure polish, and export workflows.\n\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftable>\n\n### Workspace Preview\n\n\u003Ctable>\n\u003Ctr>\n\u003Ctd width=\"33%\">\n\u003Cimg src=\"assets\u002Freadme\u002Fstart-research-promo.png\" alt=\"Start Research dialog\" width=\"100%\" \u002F>\n\u003C\u002Ftd>\n\u003Ctd width=\"33%\">\n\u003Cimg src=\"assets\u002Freadme\u002Fcanvas-promo.png\" alt=\"Canvas workspace preview\" width=\"100%\" \u002F>\n\u003C\u002Ftd>\n\u003Ctd width=\"33%\">\n\u003Cimg src=\"assets\u002Freadme\u002Fstudio-details-promo.png\" alt=\"Studio and details workspace preview\" width=\"100%\" \u002F>\n\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd valign=\"top\">\n\u003Cb>Start Research\u003C\u002Fb>\u003Cbr\u002F>\nKick off a quest from a paper, repository, or natural-language goal.\n\u003C\u002Ftd>\n\u003Ctd valign=\"top\">\n\u003Cb>Canvas\u003C\u002Fb>\u003Cbr\u002F>\nInspect branches, baselines, and accumulated research structure as a visible map.\n\u003C\u002Ftd>\n\u003Ctd valign=\"top\">\n\u003Cb>Studio + Details\u003C\u002Fb>\u003Cbr\u002F>\nReview metrics, traces, and project state without leaving the same workspace.\n\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftable>\n\n### Progress Reporting\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"assets\u002Freadme\u002Fprogress-reporting-promo.png\" alt=\"DeepScientist progress reporting example\" width=\"88%\" \u002F>\n\u003C\u002Fp>\n\n### Projects surface after long-running work\n\n![DeepScientist projects surface](assets\u002Freadme\u002Fprojects-surface.png)\n\n## Who Will Love DeepScientist Most?\n\n- graduate students and engineers who want to reproduce papers and push beyond existing baselines\n- labs or research teams running long experiment loops, ablations, and structured result analysis\n- people who want code, experiments, notes, and writing to live in one workspace\n- users who do not want to hand unpublished ideas and intermediate results directly to a pure cloud workflow\n- people who want to run work on servers while following progress from web, TUI, or messaging surfaces\n\n## The Core Philosophy Behind DeepScientist\n\nWe believe a system that is actually suitable for research should at least satisfy these principles:\n\n- one quest, one repository, instead of letting everything dissolve after a short conversation\n- branches and worktrees should express research routes naturally instead of being forced into chat history\n- failed paths should be preserved, summarized, and reused instead of overwritten\n- human researchers should always retain takeover power instead of being locked outside the loop\n- the research process should be reviewable, inspectable, and auditable instead of relying on \"the model says it did it\"\n\nIf that sounds like the way you want to work, DeepScientist is worth trying now.\n\n## 🚀 Get Started In 30 Seconds\n\nIf you want to try it right now, choose one of these two paths: run the npm commands yourself, or ask the coding tool you already use to install it for you.\n\nPlatform note: DeepScientist fully supports Linux and macOS. Native Windows support is currently experimental (strongly recommend WSL2).\n\n### Option 1: Manual Install With npm\n\nUse this path when you already know which runner you want and prefer to control the install, login, and launch commands yourself.\n\nDeepScientist ships four built-in runners:\n\n- `codex`: use this when `codex` already works directly on your machine\n- `claude`: use this when `claude` already works directly on your machine\n- `kimi`: use this when `kimi` already works directly on your machine\n- `opencode`: use this when `opencode` already works directly on your machine\n\nIf one of these CLIs already works for you, DeepScientist can usually meet you there instead of asking you to rebuild your whole setup first.\n\nThink of the startup choice like this: bring one runner that already works, and DeepScientist gives you a persistent local research workspace around it.\n\nIf you just want the safest recommendation, start with Codex first.\n\n🎯 Recommended first run: `codex`\n\n```bash\nnpm install -g @researai\u002Fdeepscientist\ncodex login\nds --here\n```\n\nIf Claude Code already works directly in your shell, use this lane:\n\n```bash\nnpm install -g @researai\u002Fdeepscientist\nclaude --version\nds doctor --runner claude\nds --here --runner claude\n```\n\nIf Kimi Code already works directly in your shell, use this lane:\n\n```bash\nnpm install -g @researai\u002Fdeepscientist\nkimi --version\nds doctor --runner kimi\nds --here --runner kimi\n```\n\nIf OpenCode already works directly in your shell, use this lane:\n\n```bash\nnpm install -g @researai\u002Fdeepscientist\nopencode --version\nds doctor --runner opencode\nds --here --runner opencode\n```\n\nIf you want to connect Gemini or Ollama, first use the runner-specific docs instead of guessing DeepScientist fields:\n\n- Gemini: prefer [OpenCode Setup](docs\u002Fen\u002F25_OPENCODE_PROVIDER_SETUP.md)\n- Ollama: choose Codex, Claude Code, or OpenCode with [Local Model Backends Guide](docs\u002Fen\u002F21_LOCAL_MODEL_BACKENDS_GUIDE.md)\n\nTo stop the managed local daemon and all currently running agents:\n\n```bash\nds --stop\n```\n\n🛠 Prefer installing from a Git checkout instead of npm? Use the repo path directly:\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002FResearAI\u002FDeepScientist.git\ncd DeepScientist\nbash install.sh\nds\n```\n\n### Option 2: Let A Coding Tool Install It\n\nUse this path when you already work inside Codex, Claude Code, OpenCode, Cursor, or another coding agent. There are only two steps:\n\n1. Launch the coding tool in a directory where you are comfortable installing DeepScientist.\n2. Copy and send this prompt:\n\n```text\nPlease install and launch DeepScientist on this machine. The official repo is https:\u002F\u002Fgithub.com\u002FResearAI\u002FDeepScientist and the docs start at https:\u002F\u002Fgithub.com\u002FResearAI\u002FDeepScientist\u002Fblob\u002Fmain\u002Fdocs\u002Fen\u002FREADME.md . First inspect Node.js\u002Fnpm, git, Python, OS, and shell environment. If global npm install is appropriate, run npm install -g @researai\u002Fdeepscientist and verify ds --help. If source install is safer, git clone https:\u002F\u002Fgithub.com\u002FResearAI\u002FDeepScientist.git, cd DeepScientist, read the README, and run bash install.sh. After installation, confirm at least one runner works locally, such as codex, claude, opencode, or kimi; authenticate that CLI first, then run ds doctor --runner \u003Cname>, start with ds --here, and report the local URL plus the exact config docs I should read next.\n```\n\nIf you plan to edit the UI or TUI from source, also install the workspace dependencies:\n\n```bash\nnpm --prefix src\u002Fui install\nnpm --prefix src\u002Ftui install\n```\n\nIf you prefer the interactive first-run flow, run this once first:\n\n```bash\ncodex\n```\n\nIf `codex` still appears to be missing after installing DeepScientist, take the explicit repair path instead of assuming the bundled dependency was linked correctly:\n\n```bash\nnpm install -g @openai\u002Fcodex\nwhich codex\ncodex login\n```\n\nIf `which codex` still prints nothing after that, fix the npm global bin path first, then retry `codex login` and `ds doctor`.\n\nImportant runner note:\n\n- DeepScientist can fall back to npm-bundled helper copies for `codex`, `claude`, and `opencode` when they are installed with the package. Kimi Code is treated as an external CLI unless a compatible local `kimi` helper is present.\n- Runner authentication and provider configuration still belong to the underlying CLI. Make `codex`, `claude`, `kimi`, or `opencode` work once in your shell, then run `ds doctor --runner \u003Cname>`.\n- You can also start DeepScientist first with the default runner and switch\u002Fconfigure Claude Code, Kimi Code, or OpenCode later from the web workspace settings.\n\nAfter startup, the default local address is:\n\n```text\nhttp:\u002F\u002F127.0.0.1:20999\n```\n\nLocal browser auth is now optional and disabled by default. If you want a per-launch local access password, start with:\n\n```bash\nds --auth true\n```\n\nThen you only need to do three things:\n\n1. click `Start Research`\n2. fill in the research goal, baseline links, paper links, or local paths\n3. let DeepScientist start a real research project that can keep evolving locally\n\nIf this is your first run, prefer an isolated environment, a non-root user, and a local machine. For the full details, see:\n\n- [00 Quick Start](docs\u002Fen\u002F00_QUICK_START.md)\n- [15 Codex Provider Setup](docs\u002Fen\u002F15_CODEX_PROVIDER_SETUP.md)\n- [24 Claude Code Setup](docs\u002Fen\u002F24_CLAUDE_CODE_PROVIDER_SETUP.md)\n- [27 Kimi Code Setup](docs\u002Fen\u002F27_KIMI_CODE_PROVIDER_SETUP.md)\n- [25 OpenCode Setup](docs\u002Fen\u002F25_OPENCODE_PROVIDER_SETUP.md)\n- [09 Doctor](docs\u002Fen\u002F09_DOCTOR.md)\n\n## 🧭 Choose Your Starting Path\n\n### ⚡ I just want to get it running first\n\n- [00 Quick Start](docs\u002Fen\u002F00_QUICK_START.md)\n- [12 Guided Workflow Tour](docs\u002Fen\u002F12_GUIDED_WORKFLOW_TOUR.md)\n\n### 🧪 I want to launch a real project today\n\n- [02 Start Research Guide](docs\u002Fen\u002F02_START_RESEARCH_GUIDE.md)\n- [01 Settings Reference](docs\u002Fen\u002F01_SETTINGS_REFERENCE.md)\n\n### 🖥 I mainly work on servers and terminals\n\n- [05 TUI Guide](docs\u002Fen\u002F05_TUI_GUIDE.md)\n  Includes `ds --tui --debug`, redacted debug JSONL, and Web\u002FTUI comparison guidance.\n\n### 🔌 I want to connect my own models or external collaboration channels\n\n- [15 Codex Provider Setup](docs\u002Fen\u002F15_CODEX_PROVIDER_SETUP.md)\n- [24 Claude Code Setup](docs\u002Fen\u002F24_CLAUDE_CODE_PROVIDER_SETUP.md)\n- [27 Kimi Code Setup](docs\u002Fen\u002F27_KIMI_CODE_PROVIDER_SETUP.md)\n- [25 OpenCode Setup](docs\u002Fen\u002F25_OPENCODE_PROVIDER_SETUP.md)\n- [21 Local Model Backends Guide](docs\u002Fen\u002F21_LOCAL_MODEL_BACKENDS_GUIDE.md)\n- [Weixin Connector Guide](docs\u002Fen\u002F10_WEIXIN_CONNECTOR_GUIDE.md)\n- [QQ Connector Guide](docs\u002Fen\u002F03_QQ_CONNECTOR_GUIDE.md)\n- [Telegram Connector Guide](docs\u002Fen\u002F16_TELEGRAM_CONNECTOR_GUIDE.md)\n- [WhatsApp Connector Guide](docs\u002Fen\u002F17_WHATSAPP_CONNECTOR_GUIDE.md)\n- [Feishu Connector Guide](docs\u002Fen\u002F18_FEISHU_CONNECTOR_GUIDE.md)\n\n### 🧠 I want to understand the system design first\n\n- [Docs Index](docs\u002Fen\u002FREADME.md)\n- [Core Architecture Guide](docs\u002Fen\u002F13_CORE_ARCHITECTURE_GUIDE.md)\n- [Prompt, Skills, and MCP Guide](docs\u002Fen\u002F14_PROMPT_SKILLS_AND_MCP_GUIDE.md)\n\n## Autonomous Research Systems\n\n### End-to-End Autonomous Research Systems\n\n| System | System Type | E2E | Research Map | Workshop | Keeps Growing | Channels | Figure & Rebuttal & Review |\n|---|---|---|---|---|---|---|---|\n| [autoresearch](https:\u002F\u002Fgithub.com\u002Fkarpathy\u002Fautoresearch) | Open-source |  |  | ✓ |  |  |  |\n| [RD-Agent](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent) | Open-source |  |  |  | ✓ |  |  |\n| [Agent Laboratory](https:\u002F\u002Fgithub.com\u002FSamuelSchmidgall\u002FAgentLaboratory) | Open-source | ✓ |  | ✓ | ✓ |  |  |\n| [AI-Scientist](https:\u002F\u002Fgithub.com\u002FSakanaAI\u002FAI-Scientist) | Open-source | ✓ |  |  |  |  |  |\n| [AI-Scientist-v2](https:\u002F\u002Fgithub.com\u002FSakanaAI\u002FAI-Scientist-v2) | Open-source | ✓ |  |  |  |  |  |\n| [AutoResearchClaw](https:\u002F\u002Fgithub.com\u002Faiming-lab\u002FAutoResearchClaw) | Open-source | ✓ |  |  | ✓ | ✓ |  |\n| [ClawPhD](https:\u002F\u002Fgithub.com\u002FZhihaoAIRobotic\u002FClawPhD) | Open-source |  |  | ✓ |  | ✓ |  |\n| [Dr. Claw](https:\u002F\u002Fgithub.com\u002FOpenLAIR\u002Fdr-claw) | Open-source | ✓ |  | ✓ |  | ✓ |  |\n| [FARS](https:\u002F\u002Fanalemma.ai\u002Ffars\u002F) | Closed-source | ✓ |  |  |  |  |  |\n| [EvoScientist](https:\u002F\u002Fgithub.com\u002FEvoScientist\u002FEvoScientist) | Open-source | ✓ |  | ✓ | ✓ | ✓ |  |\n| [ScienceClaw](https:\u002F\u002Fgithub.com\u002Fbeita6969\u002FScienceClaw) | Open-source |  |  |  | ✓ | ✓ |  |\n| [claude-scholar](https:\u002F\u002Fgithub.com\u002FGalaxy-Dawn\u002Fclaude-scholar) | Open-source | ✓ |  | ✓ | ✓ |  |  |\n| [Research-Claw](https:\u002F\u002Fgithub.com\u002Fwentorai\u002FResearch-Claw) | Open-source | ✓ |  | ✓ | ✓ | ✓ |  |\n| [DeepScientist](https:\u002F\u002Fgithub.com\u002FResearAI\u002FDeepScientist) | Open-source | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |\n\n## Documentation\n\n- [English Docs Index](docs\u002Fen\u002FREADME.md)\n- [Chinese Docs Index](docs\u002Fzh\u002FREADME.md)\n\n## NLPCC 2026 AISB Challenge\n\nIf you want to benchmark or extend AI scientist systems in the wild, the NLPCC 2026 AISB shared task is a natural next stop:\n\n- [Registration](http:\u002F\u002Ftcci.ccf.org.cn\u002Fconference\u002F2026\u002Fshared-tasks\u002F)\n- [Task Repository](https:\u002F\u002Fgithub.com\u002FResearAI\u002FNLPCC-2026-Task9-AISB)\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"assets\u002Freadme\u002Faisb-poster.jpeg\" alt=\"NLPCC 2026 AISB shared task poster\" width=\"88%\" \u002F>\n\u003C\u002Fp>\n\n## For Developers And Maintainers\n\nIf you are developing or maintaining DeepScientist, continue with:\n\n- [Architecture](docs\u002Fen\u002F90_ARCHITECTURE.md)\n- [Development Guide](docs\u002Fen\u002F91_DEVELOPMENT.md)\n- [BenchStore YAML Guide](docs\u002Fen\u002F22_BENCHSTORE_YAML_REFERENCE.md)\n- [CONTRIBUTING](CONTRIBUTING.md)\n\n## Citation\n\nIf DeepScientist materially helps your paper, report, or research workflow, please cite the DeepScientist paper and disclose meaningful AI assistance honestly.\n\nThis is a strong request for fair academic attribution, not an extra software license condition.\n\nUseful links:\n\n- Paper: `https:\u002F\u002Fopenreview.net\u002Fforum?id=cZFgsLq8Gs`\n- Repository citation metadata: [CITATION.cff](CITATION.cff)\n- Citation and attribution guidance: [docs\u002Fen\u002F26_CITATION_AND_ATTRIBUTION.md](docs\u002Fen\u002F26_CITATION_AND_ATTRIBUTION.md)\n- Acknowledgements, including optional FermiLink science-workflow attribution: [docs\u002Fen\u002F99_ACKNOWLEDGEMENTS.md](docs\u002Fen\u002F99_ACKNOWLEDGEMENTS.md)\n- Name and logo usage: [TRADEMARK.md](TRADEMARK.md)\n\nSuggested acknowledgment text:\n\n```text\nWe used DeepScientist to assist parts of the research workflow, including selected planning, implementation, experiment orchestration, analysis, and\u002For writing support. Final judgments, claims, and reported real experimental results remain the responsibility of the human authors.\n```\n\nDeepScientist is jointly developed by Yixuan Weng, Weixu Zhao, Shichen Li, Zhen Lin, and Minjun Zhu.\n\n```bibtex\n@inproceedings{\nweng2026deepscientist,\ntitle={DeepScientist: Advancing Frontier-Pushing Scientific Findings Progressively},\nauthor={Yixuan Weng and Minjun Zhu and Qiujie Xie and QiYao Sun and Zhen Lin and Sifan Liu and Yue Zhang},\nbooktitle={The Fourteenth International Conference on Learning Representations},\nyear={2026},\nurl={https:\u002F\u002Fopenreview.net\u002Fforum?id=cZFgsLq8Gs}\n}\n```\n\nIf this feels like the research workflow you have been waiting for, give the project a star. Every star makes it easier for more researchers who actually need it to find it.\n\n## Community\n\nWelcome to join the WeChat group for discussion.\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"assets\u002Freadme\u002Fwechat10.jpg\" alt=\"DeepScientist WeChat group\" width=\"360\" \u002F>\n\u003C\u002Fp>\n\n## More From ResearAI\n\nIf you like DeepScientist, you may also want to explore the rest of the ResearAI ecosystem:\n\n| Project | What it does | Stars |\n|---|---|---|\n| **[MeOS](https:\u002F\u002Fgithub.com\u002FResearAI\u002FMeOS)** | Fork yourself as a Skill, so agents understand you better | ![GitHub stars](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002FResearAI\u002FMeOS?style=flat&logo=github) |\n| [AutoFigure](https:\u002F\u002Fgithub.com\u002FResearAI\u002FAutoFigure) | generate publication-ready figures | ![GitHub stars](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002FResearAI\u002FAutoFigure?style=flat&logo=github) |\n| [AutoFigure-Edit](https:\u002F\u002Fgithub.com\u002FResearAI\u002FAutoFigure-Edit) | generate editable vector paper figures | ![GitHub stars](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002FResearAI\u002FAutoFigure-Edit?style=flat&logo=github) |\n| [DeepReviewer-v2](https:\u002F\u002Fgithub.com\u002FResearAI\u002FDeepReviewer-v2) | review papers and suggest revisions | ![GitHub stars](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002FResearAI\u002FDeepReviewer-v2?style=flat&logo=github) |\n| [Awesome-AI-Scientist](https:\u002F\u002Fgithub.com\u002FResearAI\u002FAwesome-AI-Scientist) | curated AI scientist landscape | ![GitHub stars](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002FResearAI\u002FAwesome-AI-Scientist?style=flat&logo=github) |\n\n## Roadmap\n\nWe are building DeepScientist as a long-term local-first research operating system.\n\nThe next major upgrades focus on four directions:\n\n### 1. Deeper Research Loops\n\n- AI Scientist Benchmark support for more realistic evaluation and comparison\n- smoother automatic baseline upload, download, and reuse\n- stronger experiment replay, comparison, and paper-facing outputs\n\n### 2. Stronger Long-Horizon Memory\n\n- stronger Memory and Findings Memory mechanisms\n- better cross-run and cross-quest reuse\n- less repeated failure and less rediscovery cost over long projects\n\n### 3. Richer Multimodal And Collaborative Workflows\n\n- VideoAnything-style multimodal research capabilities\n- better local-model, connector, and copilot\u002Fautonomous collaboration flows\n- a more efficient and more reliable DeepScientist system across local, collaborative, and long-horizon research settings\n\n### 4. Stronger Security And Safer Deployment\n\n- safer local-first and server-side deployment defaults\n- stronger auth, permission, and connector-surface protection\n- less fabrication, lower hallucination, and more verification-grounded outputs\n- better auditability for long-running autonomous research workflows\n\nIf this direction is interesting to you, please give the project a `Watch` and a `Star`:\n\n[![Watch DeepScientist](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fwatchers\u002FResearAI\u002FDeepScientist?style=for-the-badge&logo=github&label=Watch%20DeepScientist)](https:\u002F\u002Fgithub.com\u002FResearAI\u002FDeepScientist\u002Fwatchers)\n[![Star DeepScientist](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002FResearAI\u002FDeepScientist?style=for-the-badge&logo=github&label=Star%20DeepScientist)](https:\u002F\u002Fgithub.com\u002FResearAI\u002FDeepScientist\u002Fstargazers)\n\n---\n\nThis project is maintained by WestlakeNLP. If you run into problems, please ask on [DeepWiki](https:\u002F\u002Fdeepwiki.com\u002FResearAI\u002FDeepScientist) first; if it still cannot be resolved, open an issue.\n\nWestlakeNLP is led by ACL Fellow Professor Yue Zhang. If you are interested in a long-term internship, PhD position, or research assistant opportunity, contact Professor Yue Zhang at `zhangyue@westlake.edu.cn`.\n","DeepScientist 是一个本地优先的自主研究工作室，旨在帮助研究人员在自己的机器上完成从基线设置、实验轮次到论文准备输出的完整研究循环。该项目利用 Findings Memory、贝叶斯优化和 Research Map 等技术特性，将每个新结果转化为下一步研究的起点，支持更广泛的探索。内置了包括 Codex、Claude Code、Kimi Code 和 OpenCode 在内的多个执行器，方便用户快速启动项目。适用于需要高效管理和推进科研项目的个人或团队，特别是那些希望保持对研究过程高度控制的研究者。",2,"2026-06-11 03:47:33","high_star"]