# llama.qtcreator

A Qt Creator plugin that adds local LLM-assisted text completion and chat, backed by a locally running llama.cpp server.

llama.qtcreator adds local LLM-assisted completion to Qt Creator, backed by a llama.cpp server instance the user runs separately. It auto-suggests on cursor movement, can be triggered manually, accepts a suggestion in full or just its first line, supports speculative fill-in-the-middle and speculative decoding for speed, and builds a ring of context from open and edited files plus yanked text so large contexts work even on modest hardware.

A chat panel renders Mermaid diagrams and KaTeX math to SVG with a bundled QuickJS runtime and no browser required, supports source and image drag-and-drop, custom prompts on the current selection, and tool use. A Human Editor endpoint lets a person answer inside the chat as if they were a model, writing replies in a normal text editor pane next to the conversation.

It requires a running llama.cpp server with a fill-in-the-middle-compatible model, and is written in C++ under the MIT license.

Website: https://github.com/cristianadam/llama.qtcreator