Open-source AI research toolkit for crypto markets
Reads the market. Signals the moves. Posts the alpha.
Crypto AI Research is an open-source TypeScript research lab for crypto markets — built on Solana and powered by Claude AI. It is maintained by Crypto Signal Org.
It connects on-chain data, AI reasoning, and automated distribution into a single, modular toolkit:
- 🤖 Telegram bot that answers market questions, analyzes chart photos, and delivers signals in real time
- 📊 Signal engine with pure TypeScript TA (RSI, MACD, Bollinger Bands, EMA, ATR) — no external TA libraries
- 🐋 Whale tracker that monitors large Solana wallets and fires instant alerts on big moves
- 🐦 Auto poster that generates natural, human-style posts ready for Twitter/X
- ⚡ High-performance core written in Rust for heavy computation
Crypto AI Research doesn't predict. It researches, signals, and acts.
The CLI is published as orinlab (npm package name: orinlab).
git clone https://github.com/cryptosignalorg/crypto-ai-research.git
cd crypto-ai-research
npm install
npm run build# 1. Setup — interactive wizard, takes ~1 minute
npx orinlab setup
# 2. Run
npx orinlab bot # Telegram AI bot
npx orinlab signal # Signal engine (terminal)
npx orinlab posts SOL # Generate posts for $SOL
npx orinlab onchain # Solana wallet monitorStep 1 — AI Provider
1 Anthropic (Claude)
2 DeepInfra (free tier)
Choose provider: 2
DeepInfra API key: ••••••••
Step 2 — Telegram Bot
Telegram bot token: ••••••••
Step 3 — Solana (optional)
Configure Solana wallet tracking? [y/N]
Step 4 — Twitter/X Auto Poster (optional)
Configure Twitter/X posting? [y/N]
✓ Setup complete! Config saved to ~/.orinlab/.env
orinlab bot — start Telegram bot
orinlab signal — signal engine
Config is saved to ~/.orinlab/.env — edit anytime to update keys.
┌─────────────────────────────────────────────────────────┐
│ Crypto AI Research │
│ │
│ ┌──────────────┐ ┌──────────────┐ │
│ │ Telegram Bot │ │ Signal Engine│ │
│ │ (TypeScript) │ │ (TypeScript) │ │
│ └──────┬───────┘ └──────┬───────┘ │
│ │ │ │
│ └─────────┬─────────┘ │
│ ▼ │
│ ┌────────────────┐ │
│ │ Claude AI │ ← Anthropic / DeepInfra │
│ │ (Haiku/Sonnet)│ │
│ └────────┬───────┘ │
│ │ │
│ ┌─────────┴──────────┐ │
│ ▼ ▼ │
│ ┌─────────────┐ ┌────────────────┐ │
│ │ Auto Poster │ │ On-chain Agent│ │
│ │ (TypeScript)│ │ (TypeScript) │ │
│ └──────┬──────┘ └───────┬────────┘ │
│ │ │ │
│ ▼ ▼ │
│ Twitter/X Solana RPC │
│ (CoinGecko API) │
└─────────────────────────────────────────────────────────┘
| Module | Language | Description |
|---|---|---|
| Telegram Bot | TypeScript | AI-powered bot — market Q&A, signal delivery, chart photo analysis |
| Signal Engine | TypeScript | BUY/SELL/HOLD signals with confidence score and risk level |
| Technical Analysis | TypeScript | RSI, MACD, BB, EMA, ATR — pure TypeScript, no TA libraries |
| Market Analyst | TypeScript | Deep multi-factor market analysis using Claude Sonnet |
| On-chain Agent | TypeScript | Solana wallet monitoring, transaction parsing |
| Whale Tracker | TypeScript | Monitors large wallets, alerts on $50k+ moves |
| Post Writer | TypeScript | Natural human-style post generator for Twitter/X |
| Signal History | TypeScript | Append-only JSON signal log with stats and filters |
| Auto Poster | TypeScript | Automated signal posting to Twitter/X |
| Solana SDK | TypeScript | On-chain data fetcher — prices, wallets, transactions |
| Signal CLI | Go | Lightweight CLI for terminal-native signal checks |
| Core SDK | Rust | High-performance signal computation and data processing |
| Command | Description |
|---|---|
/start |
Welcome screen with quick-action keyboard |
/signal $TOKEN |
BUY/SELL/HOLD signal with confidence score |
/ta $TOKEN |
Full technical analysis — RSI, MACD, BB, EMA |
/analyze $TOKEN |
Deep AI market analysis (Claude Sonnet) |
/post $TOKEN |
Generate a ready-to-copy Twitter/X post |
/history |
Last 10 signals generated |
/help |
Show all commands |
Send a chart photo → instant AI chart analysis with signal, key levels, and pattern recognition.
Crypto AI Research supports multiple AI backends — switch with one env var:
| Provider | Setup | Notes |
|---|---|---|
| Anthropic | ANTHROPIC_API_KEY |
Best quality, Claude Haiku/Sonnet |
| DeepInfra | DEEPINFRA_API_KEY |
Free tier available, Llama 3.1 70B |
| OpenAI | OPENAI_API_KEY |
Any OpenAI-compatible endpoint |
| OpenRouter | OPENAI_API_KEY + OPENAI_BASE_URL |
Access to 100+ models |
# Switch provider in ~/.orinlab/.env
AI_PROVIDER=deepinfra
DEEPINFRA_API_KEY=your_key
DEEPINFRA_MODEL=meta-llama/Meta-Llama-3.1-70B-InstructClone and run locally:
git clone https://github.com/cryptosignalorg/crypto-ai-research.git
cd crypto-ai-research
npm install
cp .env.example .env # fill in your keys
npm run build
npx orinlab signalnpm test
npm run test:pipelinecrypto-ai-research/
├── src/ ← TypeScript core (agents, bot, CLI, utils)
│ ├── agents/ ← signal engine, TA, whale tracker, etc.
│ ├── bot/ ← Telegram bot
│ ├── cli/ ← orinlab CLI entry point
│ └── redis/ ← cache layer
├── poster/ ← TypeScript auto poster
├── sdk/ ← TypeScript Solana SDK
├── cli/ ← Go signal CLI
├── core/ ← Rust core SDK
├── tests/ ← Vitest unit tests
├── scripts/ ← pipeline smoke tests
└── .github/ ← CI/CD, CodeQL, issue templates
Contributions are welcome. Read CONTRIBUTING.md first.
- Fork the repo
- Create a branch:
git checkout -b feature/your-feature - Follow Conventional Commits
- Open a PR against
main
Found a vulnerability? See SECURITY.md. Do not open a public issue.
Crypto AI Research is experimental software for research purposes.
- AI signals are not financial advice
- Always DYOR
- Never share your private key
- Use at your own risk
MIT — see LICENSE