hikwoa-hfw/kaffra — reverse-engineered prompt
Reverse engineered prompt
Build me a Telegram bot called Charon that watches a private Pump.fun signal feed and helps me screen noisy token launches before buying.
I want it to poll a signal server with an API key, filter tokens by things like source overlap, fees, token age, market cap, holders, trend quality, ATH distance, and position limits, then enrich good candidates with Jupiter, GMGN if enabled, saved wallet exposure, chart or social context when available. It should optionally ask an OpenAI compatible LLM to pick the best candidate and explain why.
The bot should work from Telegram with a menu and commands for strategies, filters, positions, PnL, saved wallets, candidates, manual sells, and learning notes. It needs dry run mode with fake trades in SQLite, confirm mode with approve or reject buttons, and live mode using Solana and Jupiter swaps when keys are provided.
Please include configurable strategies like sniper, dip buy, smart money, degen, and profit lock, with hot editable settings stored in SQLite. Open positions should survive restart and be monitored for take profit, stop loss, trailing, partial TP, max hold, and profit lock rules.
Want more depth? Deep Reverse