schema_version = 1

extends     = "llama.cpp Base Chat"
name        = "llama.cpp Chat"
description = "Local llama.cpp (llama-server) — coding chat via the OpenAI-compatible Chat Completions API. llama-server serves whichever GGUF you loaded, so 'model' is only a label. Tool calling needs a tool-capable model and llama-server started with --jinja."

model        = "qwen2.5-coder-7b-instruct"
enable_tools = true
tags         = ["chat", "llama.cpp", "local"]

system_prompt = """{{ read_file(":/roles/qt-cpp-developer.md") }}"""

[body]
max_tokens  = 8192
temperature = 0.7