grihithvasishta/VENOM — reverse-engineered prompt
Reverse engineered prompt
Build me a terminal app called VENOM, a TypeScript CLI that feels like a smart AI coding assistant for the command line.
I want it to use multiple AI models together instead of just one. For normal prompts, send the question to the available agents at the same time, then have a main orchestrator combine the answers fairly, call out useful differences, and give one strong final response. For coding requests, make it follow a pipeline where one model plans, one researches best practices, and one writes the code.
Please include an interactive REPL, a one shot mode, and these slash commands: code, review, debug, explain, refactor, plan, research, test, commit, and help. It should have a nice terminal experience with markdown and code rendering, loading states, saved conversation history, and automatic project context by reading things like package.json, config files, and the file tree.
Also add setup for API keys, a local config file, and a Telegram remote mode so I can send prompts from Telegram and get replies back. If you need details, look up current docs online.
Want more depth? Deep Reverse