agentcoinorg/evo.ninja — reverse-engineered prompt
Reverse engineered prompt
Build me a local AI assistant called evo.ninja that can take a goal from me and keep working on it in a loop until it feels done. I want it to act like a generalist agent that can switch between a few built in roles depending on the job, like reading and summarizing text files, analyzing CSV data, researching things on the web, and helping write software. It should choose the best role for each step, use only the most relevant chat history for that step, and then execute the next action.
Please make it usable both from the terminal and from a simple browser UI. In the terminal, I should be able to start a named session, give it a timeout, and optionally save debug logs. Each session should get its own workspace folder where the agent can read and write files, plus saved chat output and internal logs.
Use environment variables for the AI key and search key, and set up the local web app so it can run with the needed local database services too. Look up current docs online if you need to.
Want more depth? Deep Reverse