HanSyngha/Local-CLI — reverse-engineered prompt

Build me a local first AI coding assistant that works both as a terminal tool and a simple Windows desktop app. I want to point it at my own OpenAI compatible model endpoint, like Ollama, vLLM, LM Studio, Azure OpenAI, or anything similar, so it can run fully inside a company network with no cloud dependency.

It should be able to chat, read and edit project files, search through code, and handle bigger requests by making a step by step plan and then executing it. Please make it safe by default, with a supervised mode where it asks before changing files. I also want a desktop experience with a chat window and a separate always on top task window so I can watch progress in real time.

If possible, include pipe mode for automation, image input for vision models, browser driven research with citations, and Windows office automation for Word, Excel, and PowerPoint. Confluence and Jira support through a visible browser would be great too. Look up current docs online if you need to.