ANARCHY is a newly released, fully local AI shell designed for developers who prioritize privacy and offline functionality. It leverages Ollama as its primary LLM engine, with GGUF models as a fallback, ensuring all AI processing occurs on local hardware without requiring cloud access or API keys. The shell supports a wide range of operating systems, including Windows, macOS, Linux, Chromebook, and Android (via Termux).
The v1.0.0 release introduces several key features. A core engine with 40 subsystems includes an NLP/NLU pipeline for intent classification and entity extraction, a tool execution engine with tiered permissions, and semantic memory using FAISS and JSON. It also features smart model routing to select appropriate model sizes based on query complexity, and RAG search for indexing and querying codebases. Voice input is handled locally via Whisper transcription, and a cron-style task scheduler allows for automated LLM prompts.
Notable new additions in this release include LSP Code Intelligence, which runs local language servers (e.g., pyright, clangd) to provide static analysis errors explained by the local LLM. A 'Smart Git Workflow' automates commit message drafting based on staged diffs and repository style, and facilitates `commit → push → PR` in a single command. ANARCHY also includes a Git Hook Manager and built-in review skills for security and code auditing, invoked through natural language.