#5959th of 92 in RAG, knowledge and searchFor science, health and research, research and papers

deep-research-web-ui

Web UI that turns a question into a cited research report

Releases and problems feed (RSS)

Stars
2.2k
License
no license
Last commit
Nov 2026
Last release
Sep 2026
Language
TypeScript
Stars, last 29 weeks+90
30 Apr 2026: 2,138 stars7 May 2026: 2,142 stars14 May 2026: 2,144 stars21 May 2026: 2,146 stars28 May 2026: 2,149 stars4 Jun 2026: 2,152 stars11 Jun 2026: 2,153 stars18 Jun 2026: 2,158 stars25 Jun 2026: 2,161 stars2 Jul 2026: 2,163 stars9 Jul 2026: 2,164 stars16 Jul 2026: 2,170 stars23 Jul 2026: 2,173 stars30 Jul 2026: 2,179 stars6 Aug 2026: 2,182 stars13 Aug 2026: 2,187 stars20 Aug 2026: 2,189 stars27 Aug 2026: 2,190 stars3 Sept 2026: 2,190 stars10 Sept 2026: 2,194 stars17 Sept 2026: 2,200 stars24 Sept 2026: 2,208 stars1 Oct 2026: 2,212 stars8 Oct 2026: 2,216 stars15 Oct 2026: 2,219 stars22 Oct 2026: 2,219 stars29 Oct 2026: 2,221 stars5 Nov 2026: 2,223 stars12 Nov 2026: 2,226 stars19 Nov 2026: 2,228 stars

2,138 on 30 Apr 20262,228 on 19 Nov 2026

From GitHub's daily star counts.

Checked on GitHub.

Overview

Nuxt app that has an LLM plan web searches, reads source pages, and writes a cited report alongside a search tree. Citations open an evidence panel, and follow-ups update the affected passages as a new history entry. It runs in Client Mode (keys in the browser) or Server Mode (keys in env vars, Docker image on port 3000).

Who it is for: Self-hosters wanting a browser-based deep research tool with their own API keys

Strengths

  • Uses plain prompts, so OpenAI-compatible, Ollama, LiteLLM and DeepSeek endpoints work
  • Citation inspection with matched source excerpts; follow-ups keep the original report in history
  • Client Mode keeps keys and requests in the browser and can be deployed statically
  • Server Mode adds access password, search key rotation and HTTP/SOCKS outbound proxy

Weaknesses

  • Needs an external web search provider such as Tavily, Firecrawl or Google PSE
  • History is stored only in the browser; export is manual
  • Public Server Mode exposes your keys unless NUXT_ACCESS_PASSWORD is set
  • SOCKS proxies do not cover the Tavily, Firecrawl and CRW SDKs

What it needs

  • no GPU
  • Docker
  • Needs LLM provider API, Web search API (Tavily, Firecrawl, fastCRW, Google PSE, You.com, Serply or Parallel)
  • Models: OpenAI-compatible, SiliconFlow, DeepSeek, OpenRouter, Requesty
  • port 3000

What changed

From the release notes. No flag means the note does not say so, not that the upgrade is safe.

Also in RAG, knowledge and search

See all 92
Also in RAG, knowledge and search
RankProjectScore
1Open NotebookSelf-hosted NotebookLM alternative for research notebooks, chat, and podcasts42.1k starsMIT93 out of 100
2LightRAGGraph-based RAG framework with a REST server and WebUI40.8k starsMIT93 out of 100
3MaxKBEnterprise knowledge-base agent platform with RAG, workflows and MCP tools23.2k starsGPL-3.092 out of 100
#4RAGFlowRAG engine with document parsing, agentic retrieval and citations Live demo ↗ (opens in a new tab)94.5k starsApache-2.090 out of 100
#5deeptutorTutoring workspace with knowledge bases, guided learning, memory and quizzes61.2k stars+19,292 (+46%) in 30 daysApache-2.090 out of 100