#3737th of 42 in RAG, knowledge and search
deep-research
Iterative deep-research agent that searches, reads and writes a markdown report
- Stars
- 19.9k
- License
- MIT
- Last commit
- Apr 2026
- Language
- TypeScript
18,467 on 4 Apr 202619,868 on 31 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
A TypeScript CLI that takes a query, asks follow-up questions, then runs recursive search rounds through Firecrawl and an LLM, controlled by breadth and depth parameters. Each round extracts learnings and new research directions, and the result is written to report.md or answer.md. The code is kept deliberately small (target under 500 lines) so it can be read and extended.
Who it is for: Developers who want a minimal, hackable deep-research agent in Node.js
Strengths
- Small codebase, targeted at under 500 lines, easy to read and fork
- Breadth and depth parameters give direct control over research scope
- Works with OpenAI-compatible endpoints, local servers, or Fireworks DeepSeek R1
- Supports self-hosted Firecrawl via FIRECRAWL_BASE_URL
Weaknesses
- Requires a Firecrawl key or self-hosted Firecrawl for search and scraping
- CLI only; no web UI or API server described
- Free Firecrawl tier can hit rate limits, forcing concurrency down to 1
- No tagged releases; Docker setup steps are manual
What it needs
- no GPU
- Docker + Compose
- Needs Firecrawl API or self-hosted Firecrawl, OpenAI API key or OpenAI-compatible endpoint, Node.js
- Models: OpenAI o3-mini, DeepSeek R1 via Fireworks, OpenAI-compatible endpoints (OpenRouter, Gemini, local servers)
Quick answers
Is deep-research open source?
Yes. Its code is on GitHub under the MIT license.
Can you run deep-research with Docker?
Yes. Its repository has a Docker setup: a Dockerfile, a Compose file or a published image.
Does deep-research need a GPU?
No, it runs without one.
Also in RAG, knowledge and search
See all 42| Rank | Project | Score |
|---|---|---|
| 1 | RAGFlowRAG engine with deep document parsing, template chunking and agentic retrieval | 90 out of 100 |
| 2 | LightRAGGraph-based RAG framework with a REST API server and web UI | 89 out of 100 |
| 3 | deeptutorSelf-hosted AI tutor with knowledge bases, guided learning, books and quizzes | 88 out of 100 |
| #4 | MaxKBEnterprise knowledge-base agent platform with RAG, workflows and MCP tools | 86 out of 100 |
| #5 | quivrGo engine that ingests content streams for hybrid search and keyword alerts | 84 out of 100 |