Golems - Autonomous AI Agents
A monorepo ecosystem for autonomous AI agents. Features include: autonomous coding loops with PRD-driven development (Ralph), BrainLayer persistent memory with semantic search using sqlite-vec and sentence-transformers, Telegram bot integration for remote control and notifications, Night Shift for automated improvements while sleeping, and multi-agent sprint coordination via OrcClaude v2.0.
golems
0
Packages
0
Skills
0
Supported CLIs
0+
PRs merged
Built with
Key features
AI-Agnostic Skills
55 skills with 3-layer architecture: SKILL.md (universal) + adapters/ (per-CLI) + capabilities.yaml (routing). Validated across Claude, Codex, Cursor, Gemini, and Kiro.
Eval Framework
40 skill eval packs with 480+ assertions and fixture-based testing. 96% pass rate. Skills are measured, not assumed to work.
OrcClaude v2.0
Orchestrator agent with planning topology, response markers, and multi-agent sprint coordination. Spawns and monitors parallel Claude workers.
How it works
Telegram
User commands
OrcClaude v2
Multi-agent sprints
55 Skills
5 CLI adapters
ClaudeCodexCursorGeminiKiro
Eval Framework
480+ assertions
Cloud + Local
Railway + Mac
Telegram
User commands
OrcClaude v2
Multi-agent sprints
55 Skills
5 CLI adapters
ClaudeCodexCursorGeminiKiro
Eval Framework
480+ assertions
Cloud + Local
Railway + Mac
Get started
git clone https://github.com/EtanHey/golems.git
cd golems && bun installThe journey
01
Monolith to Monorepo
Started as a single-file Telegram bot. Grew into an autonomous agent ecosystem. Componentized through a 9-phase plan into 10 Bun workspace packages — each a self-contained Claude Code plugin.
02
7 Domain Golems
Each golem owns a domain: Claude (orchestrator + Telegram), Jobs (scraping + matching), Recruiter (outreach + practice), Teller (finance), Content (visual + text), Coach (calendar + planning), Services (Night Shift + briefing + cloud worker).
03
Cloud/Local Split
Mac = Brain (Telegram bot, Night Shift, local LLMs). Railway = Body (email poller, job scraper, morning briefing). Multi-LLM routing: Gemini Flash-Lite (free cloud) → GLM-4.7-Flash (free local) → Haiku (paid fallback).
04
MCP Server Ecosystem
8 MCP servers exposing 60+ tools: BrainLayer (memory), VoiceLayer (voice I/O), email triage, job discovery, GLM summarization, Supabase access, web search, and browser automation.
05
Autonomous Coding Loop
Ralph executes PRD stories autonomously: read story → implement → CodeRabbit review → fix → commit. Night Shift runs at 4am scanning repos for TODOs, creating worktrees, and shipping improvements while the human sleeps.
06
Memory + Voice Integration
BrainLayer gives agents persistent memory. VoiceLayer gives them voice. Together with Golems orchestration: agents that remember, speak, listen, and act autonomously across every project.
Part of the ecosystem
BrainLayer — Persistent Memory for AI Agents
11 MCP tools, 312K+ indexed chunks, hybrid semantic+keyword search, knowledge graph with entity resolution, local LLM enrichment via Groq/MLX/Ollama. pip install brainlayer.
VoiceLayer — Voice I/O for AI Coding Agents
2 MCP tools (voice_speak + voice_ask), 5 voice modes, whisper.cpp STT (~300ms), edge-tts, VoiceBar MCP daemon, session booking. 314 tests. bunx voicelayer-mcp.