Step-by-step guides to master every niuma_code feature, from quick start to advanced capabilities.
Download, configure, and launch niuma_code in under 5 minutes
Let LLM explore autonomously in a tool loop, thinking and doing simultaneously
Goal-driven self-loop with plan → execute → verify → self-correct
Full-screen interactive interface with 5 modal overlays
Full-screen editor with Python-native orchestration scripts
Configure multiple API endpoints, auto-route by model name
N parallel contexts with LRU eviction and visual switching
Built-in LanguageParser symbol graph with 4 retrieval tools (Python, JS, TS, Go, Java, Rust, Ruby, PHP)
Persistent memory from 10 real-time perception events
Isolated read-only sub-agents for parallel tool execution