ai-research-agent

Open-source deep research MCP. Qwen3-30B-A3B-Thinking via OpenRouter, cited web synthesis for Claude Code, Codex, Cursor, Hermes and any MCP-compatible agent.

Personalized top-conference recommendations, human-guided paper notes, and shareable research galleries that grow with your work.

Korea-market multi-agent equity research framework (KOSPI/KOSDAQ) — CLI-first LLM execution over KIS/DART/ECOS/Naver News with a self-improving analyze→verify→reflect loop. Research/education only.

Autonomous autoresearch framework for Codex and Claude Code, combining experiment DAGs, scientific memory,literature grounding, agentic planning, reproducible benchmarks, and local dashboards.

AI agent for market and audience research: web research with citations, document ingestion for RAG, audience/persona profiling — backed by real web-search and document APIs

kuki search is a fast and concurent reserch engine for deep web analysis and synthesis. Part of kuki ai agent (works standalone too)

Claude Code Plugin with AI research agent skills powered by MCP. 10 MCP tools, 5 agent skills, 6 AI research agents via RivalSearchMCP. Zero API keys.

Self-correcting AI research agent built with LangGraph, Tavily Search, Ollama, FastAPI, and React. Decomposes complex questions, retrieves web evidence, verifies relevance, and generates citation-grounded answers.

Fused AutoResearch: Karpathy Autoresearch + Anthropic / OpenAI / Martin Fowler Harness 深度融合的生产级长期自主 AI 编码研究员系统。人类只写 program.md，AI 代理 24/7 自主研究、迭代、实验、自维护。

A comprehensive MVP for NYX Cipher—a decentralized AI research agent designed for crypto trading, influencer tracking, and automated reporting.

Multi-agent AI research pipeline (Search → Read → Write → Critique) built with LangChain and Groq. Autonomously finds live sources on a topic, scrapes the most relevant one, drafts a structured report, and self-critiques it against a scoring rubric.

Zero-cloud, privacy-preserving research intelligence platform with automated web extraction, academic report synthesis, and dense vector RAG.

Self-correcting multi-agent research engine for technical briefs. Planner drafts queries, Tavily retrieves in parallel, a critic grades 1-10 and loops until >=8 or max depth, then a synthesizer writes Markdown. Next.js 14, LangGraph.js, Gemini 2.5 Flash, SSE. Public Vercel is Demo mode (simulated). Live needs GEMINI_API_KEY and TAVILY_API_KEY. MIT.