AICH — cryptocurrency research workspace

A usable local first slice of the original AI-powered Cryptocurrency Hub concept: a research dashboard, transparent risk metrics, methodology notes and a persistent research forum.

Run

Python 3.10+; no third-party dependencies, API keys or network required.

python app.py
# Open http://127.0.0.1:8080
python -m unittest discover -s tests -v

Choose BTC, ETH or SOL to explore a 90-day price chart, total return, annualized sample volatility and maximum drawdown. Add a research note; it survives server restarts in workspace.sqlite3. Set --db /path/to/file.sqlite3 for another workspace or --port 9000 for another port.

What the demo actually does

The dashboard reads the same HTTP API a client can use:

curl 'http://127.0.0.1:8080/api/analysis?symbol=BTC'
curl http://127.0.0.1:8080/api/news
curl -H 'Content-Type: application/json' -d '{"author":"Researcher","body":"Compare volatility with drawdown."}' http://127.0.0.1:8080/api/posts
curl http://127.0.0.1:8080/api/posts

The API validates text limits, persists notes with parameterized SQLite statements, returns JSON errors, blocks cross-origin browser writes and serves only allowlisted static files. User text is rendered as text, never HTML.

All prices and methodology notes are explicitly synthetic fixtures, dated January 1, 2025. Risk metrics are calculated from those fixtures. There is no live market feed, LLM prediction model, scraped news, wallet or trading path. This is a local research tool, not an authenticated multi-user service. It binds to loopback only; do not expose it publicly without authentication, limits, CSRF protection and a production HTTP server.

Remaining product work

Before calling this an AI or live-market product, select a licensed market/news provider and verify timestamps, missing prices and rate-limit behavior. Any model needs a defined prediction target, chronological evaluation, leakage checks and fee-aware backtests. Multi-user hosting needs identity, scoped data access and deployment tests. The original README's NestJS/Next.js tree was a plan rather than existing code; this slice deliberately establishes runnable behavior before introducing those dependencies.

Recording walkthrough

Start with a clean --db /tmp/aich-demo.sqlite3, open the dashboard, switch assets, expand the methodology, save a note and reload to show persistence. Keep the synthetic-data badge visible. Record only after the tests and browser workflow pass at the exact recorded commit.