# BACH - Text-Based Operating System for LLMs

> BACH is a text-based operating system that empowers Large Language Models to work autonomously, learn, and self-organize. Part of the ellmos family (Extra Large Language Model Operating Systems), BACH provides infrastructure for task management, structured memory, multi-agent orchestration, and automation -- not a framework like LangChain or LlamaIndex, but the working environment underneath.

BACH differs from agent frameworks (LangChain, LlamaIndex, Haystack) by treating the LLM as a resident of a persistent text-based OS rather than a caller of a pipeline. Core primitives: handlers (CLI/API surface), skills (reusable workflows), chains (llmauto prompt sequences), boss-expert agents, a SharedMemory Bus for cross-agent coordination, and a scheduler for time-based and event-driven automation. Memory is first-class: six memory types with decay, conflict detection, and consolidation across 210+ DB tables.

Current version: v3.9.1-tiramisu. Scale: 113+ handlers, 550+ tools, 1870+ skills, 58 workflow templates. Stack: Python 3.10+, SQLite. License: MIT.

## Docs

- [README](https://github.com/ellmos-ai/bach/blob/main/README.md): Project overview, installation, key features, ellmos family positioning
- [README (Deutsch)](https://github.com/ellmos-ai/bach/blob/main/README.de.md): Full German documentation
- [Quickstart Guide](https://github.com/ellmos-ai/bach/blob/main/QUICKSTART.md): First workflow in 5 minutes
- [User Manual](https://github.com/ellmos-ai/bach/blob/main/BACH_USER_MANUAL.md): Complete handbook, all subsystems
- [SKILL.md](https://github.com/ellmos-ai/bach/blob/main/SKILL.md): LLM operating instructions (Claude, Gemini, Ollama)
- [CHANGELOG](https://github.com/ellmos-ai/bach/blob/main/CHANGELOG.md): Version history and breaking changes

## Core subsystems

- [Agents Catalog](https://github.com/ellmos-ai/bach/blob/main/AGENTS.md): Boss-expert orchestration model, 11 boss agents + 22 experts
- [Skills Catalog](https://github.com/ellmos-ai/bach/blob/main/SKILLS.md): All 1870+ reusable workflows and templates
- [Workflows](https://github.com/ellmos-ai/bach/blob/main/WORKFLOWS.md): 58 workflow templates (pre-built processes)
- [Chains](https://github.com/ellmos-ai/bach/blob/main/CHAINS.md): llmauto chain steps with `bach://` URL resolution
- [Memory System](https://github.com/ellmos-ai/bach/blob/main/MEMORY.md): Six memory types, decay, consolidation, conflict detection
- [Ollama Integration](https://github.com/ellmos-ai/bach/blob/main/OLLAMA.md): Local LLM backend configuration
- [Contributing Guide](https://github.com/ellmos-ai/bach/blob/main/CONTRIBUTING.md): Contribution workflow

## ellmos family (related projects)

- [USMC](https://github.com/ellmos-ai/usmc): Tier 1 -- United Shared Memory Client, the spring (shared memory primitive only)
- [Rinnsal](https://github.com/ellmos-ai/rinnsal): Tier 2 -- the trickle, USMC + llmauto (LLM orchestration), extremely compact
- [BACH](https://github.com/ellmos-ai/bach): Tier 3 -- the stream that unites everything, full OS with handlers, agents, GUI, bridges

## MCP servers (IDE integration)

- [ellmos-codecommander-mcp](https://github.com/ellmos-ai/ellmos-codecommander-mcp): Code analysis, JSON repair, encoding fix, import organization (14 tools)
- [ellmos-filecommander-mcp](https://github.com/ellmos-ai/ellmos-filecommander-mcp): File management and batch operations

## Optional

- [GitHub Issues](https://github.com/ellmos-ai/bach/issues): Bug reports and feature requests
- [GitHub Discussions](https://github.com/ellmos-ai/bach/discussions): Community Q&A
- [Code of Conduct](https://github.com/ellmos-ai/bach/blob/main/CODE_OF_CONDUCT.md): Project community standards
- [License (MIT)](https://github.com/ellmos-ai/bach/blob/main/LICENSE): License text
