Telegram Crypto Risk Bot MVP
Budget: €250 – €750 EUR
I need a working proof-of-concept that lets anyone drop a wallet address or transaction hash into a Telegram chat and instantly receive a plain-English risk report. The flow is straightforward:
1. The user presses an inline button or picks from a simple menu, then submits an Ethereum, Binance Smart Chain, or Polygon address / tx hash.
2. The backend calls the appropriate on-chain API (Etherscan, BscScan, Polygonscan or a comparable JSON-RPC endpoint) to pull balances, age, token flows, contract flags, and any black-list data you find relevant.
3. Those raw metrics are fed into a pre-trained language model that returns a three-level rating—Low, Medium, or High risk—followed by a short explanation and a practical recommendation.
4. The bot posts that summary back to the chat in a compact, human-friendly format.
I am open on tooling, but Python (aiogram / python-telegram-bot) or Node.js (Telegraf) plus an LLM wrapper such as OpenAI’s GPT or Cohere would be ideal; use whatever gets the MVP shipped fastest. Hosting can sit on Heroku, Render, or a small VPS—include simple instructions so I can redeploy.
Deliverables:
• Telegram bot source with inline buttons and a /menu command.
• Backend service that fetches chain data for ETH, BSC, and Polygon and calls the chosen LLM.
• Environment file or Docker setup with all keys and endpoints abstracted.
• Quick README covering setup, secrets, and how to extend the risk logic.
Acceptance tests:
• Sending a valid address on each chain produces a report in <5 s.
• Submitting gibberish triggers a graceful validation error.
• L, M, H labels correspond to threshold values I can tweak in a config file.
• The bot continues to respond after a reload with no manual steps.
If this pilot runs smoothly I plan to expand into scoring NFTs, DeFi pools, and cross-chain bridges, so clean, modular code is important. Let me know which stack you prefer and how quickly you can demo a first version.
1. The user presses an inline button or picks from a simple menu, then submits an Ethereum, Binance Smart Chain, or Polygon address / tx hash.
2. The backend calls the appropriate on-chain API (Etherscan, BscScan, Polygonscan or a comparable JSON-RPC endpoint) to pull balances, age, token flows, contract flags, and any black-list data you find relevant.
3. Those raw metrics are fed into a pre-trained language model that returns a three-level rating—Low, Medium, or High risk—followed by a short explanation and a practical recommendation.
4. The bot posts that summary back to the chat in a compact, human-friendly format.
I am open on tooling, but Python (aiogram / python-telegram-bot) or Node.js (Telegraf) plus an LLM wrapper such as OpenAI’s GPT or Cohere would be ideal; use whatever gets the MVP shipped fastest. Hosting can sit on Heroku, Render, or a small VPS—include simple instructions so I can redeploy.
Deliverables:
• Telegram bot source with inline buttons and a /menu command.
• Backend service that fetches chain data for ETH, BSC, and Polygon and calls the chosen LLM.
• Environment file or Docker setup with all keys and endpoints abstracted.
• Quick README covering setup, secrets, and how to extend the risk logic.
Acceptance tests:
• Sending a valid address on each chain produces a report in <5 s.
• Submitting gibberish triggers a graceful validation error.
• L, M, H labels correspond to threshold values I can tweak in a config file.
• The bot continues to respond after a reload with no manual steps.
If this pilot runs smoothly I plan to expand into scoring NFTs, DeFi pools, and cross-chain bridges, so clean, modular code is important. Let me know which stack you prefer and how quickly you can demo a first version.