assafelovic/gpt-researcher
An autonomous agent that conducts deep research on any data using any LLM providers
26k stars, autonomous research agent, and CT hasn't found it yet.
Turn what you learned into a concrete stack decision.
Want the shortlist in your inbox?
Subscribe for the weekly brief that turns new AI noise into the few tools and workflows worth testing.
An autonomous agent that conducts deep research on any data using any LLM providers
Guide
5 AI Repos That Mattered This Week (April 18)
Three MCP servers, a desktop agent, and a framework. The protocol layer is winning.
Guide
How to Build an AI Agent That Makes Money
From toy project to revenue — practical monetization patterns for autonomous AI agents
Guide
Maigret: The OSINT Tool Your Research Agent Is Missing
Query 3,000+ sites from one username. Build a full dossier in seconds.
Most DeFi due diligence looks like this: skim a few tweets, ctrl+F the docs, ask Perplexity a vague question, call it research. That's fine until you're deploying real capital into a protocol you've spent 20 minutes reading about.
gpt-researcher is an open-source autonomous research agent that does the deep part for you — multi-source, parallel search, structured report output, and it works with any LLM backend you point it at. 26k GitHub stars. Almost nobody in crypto Twitter is actually running it. That gap is worth paying attention to.
You give it a research question. It runs an autonomous agent loop: generates sub-queries, searches multiple sources in parallel, scrapes and synthesizes the results, then writes a structured report. Not a summary — a full research document with citations.
The key difference from asking Perplexity or ChatGPT: it doesn't stop at the first plausible answer. It runs multiple searches, cross-references sources, and flags contradictions. For protocol research, that matters. Perplexity will confidently tell you a TVL number that's 4 months stale. gpt-researcher will pull from docs, forums, audits, and news simultaneously.
It's also local. You're not piping your alpha into someone's SaaS. Run it on your machine, point it at your API keys, keep your research private.
Protocol vetting before entry. Drop in a question like "What are the known risks, audit history, and tokenomics of [protocol]?" and get a 2,000-word report covering smart contract audits, team background, tokenomics critiques from forums, and recent governance activity. Things you'd spend an hour piecing together manually.
Comparing yield strategies. "Compare staking APY, lockup terms, and smart contract risk between Protocol A and Protocol B." It'll pull from docs, DeFi aggregators, and community discussions at the same time.
Tracking narrative shifts. Research agents are good at synthesizing scattered signals. "What is the current sentiment around real-world asset protocols in the last 30 days?" pulls from news, crypto forums, and social sources into one coherent picture. Useful for positioning before a narrative pops, not after.
Governance research. Voting on a proposal? Give it the proposal text and ask it to research the team's track record, similar proposals in the space, and community arguments for and against. You'll read faster than 30 minutes of Discord archaeology.
Install is straightforward if you're comfortable with a terminal. If you're not, there's a Docker path that's basically copy-paste.
pip install gpt-researcher
Set your API keys (OpenAI by default, but you can swap in Anthropic, Groq, Ollama, etc.), then run:
from gpt_researcher import GPTResearcher
researcher = GPTResearcher(
query="What are the security risks and audit history of Aave v3?",
report_type="research_report"
)
report = await researcher.conduct_research()
output = await researcher.write_report()
It returns a formatted markdown report you can save, share, or pipe into a note.
The web UI is also available if you'd rather not touch code. Docker spins it up in a few minutes.
Full details on the tool page.
Strong: Speed vs. depth tradeoff is genuinely good. What takes a careful analyst 45-60 minutes takes gpt-researcher 3-5 minutes. For initial vetting — deciding whether something deserves more attention — that's valuable. The multi-source parallel search is the real differentiator. Single-source AI tools miss context that exists in forums, GitHub issues, or governance discussions.
Falls short: It's only as good as what's indexed. For obscure protocols, low-liquidity chains, or anything that lives mostly in Telegram or private Discord, the web search layer won't capture it. You still need your information edges — the stuff that isn't public.
Also: it won't tell you what to do with the information. Research output, not signal generation. Don't confuse a well-cited report for a trade thesis.
Cost: Runs on your API keys. A deep research report typically costs $0.05–0.30 in API calls depending on depth and model. At that price you can run it on every protocol you're evaluating and it still costs less than one bad position.
CT has adopted AI for content, for chatbots, for trading bots. Research is the last manual workflow. The traders doing systematic protocol research with tools like this have a real information processing advantage — not because AI is smarter, but because they're covering 10x more ground in the same time.
The gap won't stay open. It rarely does once something crosses 20k stars.
Written by McKlaud AI. Want to know which AI tools actually fit your business? Get a free AI audit.