# gleann

A single-binary, Go implementation of a RAG workspace and autonomous agent framework, inspired by the Leann backend, that runs embeddings and inference locally through Ollama or llama.cpp with AST-aware code indexing.

gleann is a terminal-oriented RAG workspace and agent framework written in Go, built as a self-contained alternative to Leann's Python-based architecture: every embedding and inference call runs locally through Ollama or llama.cpp, and the whole system installs as a single binary with no Python virtual environment, Node.js runtime or Docker required. A single install command wires the resulting knowledge base into seventeen AI editors and agents, including Claude Code, Cursor, Windsurf, Zed, JetBrains, Neovim, Codex, Gemini CLI and GitHub Copilot CLI.

Beyond plain text search, it performs AST-aware chunking, call-graph traversal and blast-radius analysis so that an LLM receives structural code context rather than only nearby text, and its indexes live in a single portable directory that can be copied between machines with the whole knowledge base intact. Its agent layer follows a reasoning-and-acting pattern with direct LLM integration, and the project credits the Leann project's selective-recomputation approach as the basis for its retrieval design, while reimplementing it as a compact, dependency-free Go binary.

Website: https://github.com/tevfik/gleann