Crypto Reliability Tracker

• 

  Homepage

• 

  Score report / Search result

• 

  Workflow

Inspiration

Crypto research is noisy and fragmented, price stats, repo activity, and on-chain signals live in different places. We wanted one place that turns those signals into an explainable reliability score.

What it does

Users type a coin name and get a Master Score, Confidence, three subscores (Market Integrity, Dev Velocity, On-chain Security)

How we built it

A Next.js UI calls an internal /api/score route, which spawns a Python orchestrator. The orchestrator runs a Node script to fetch CoinGecko + optional Alchemy context, cleans it into a strict JSON schema, then sends it to three Backboard agents that return JSON subscores. The orchestrator combines them via a weighted average and returns one final JSON payload to the frontend.

Challenges we ran into

We fought a lot of “plumbing” issues: mixed logs breaking JSON parsing, truncated stdout, prompt-template {} conflicts, non-EVM “wrapped” addresses causing provider errors, and Next.js App Router async pitfalls (e.g., searchParams/headers() needing async handling).

Accomplishments that we're proud of

We shipped an end-to-end pipeline that reliably returns structured, explainable scores in seconds, with clean separation between data collection, scoring, and presentation.

What we learned

Trust comes from contracts and explainability: strict JSON outputs, normalized confidence, clear flags, and disciplined logging matter as much as the model call.

What's next for Crypto Reliability Tracker

Add Social Sentiment, improve EVM/non-EVM detection, cache results to reduce API rate pressure, and refine scoring calibration with benchmarks against known scam/legit token lists.

Built With

• backboard.io
• javascript
• nextjs
• python
• typescript