Gen Code — the AI agent
that lives in your terminal

Built on five pluggable pillars — LLMs, search, personas, skills & extensions, and a self-evolving agent that levels up as you work. One Go binary; a unified runtime for specialized agents.

Release Stars Go Report Card License
$ curl -fsSL https://raw.githubusercontent.com/genai-io/gen-code/main/install.sh | bash
Get started → View on GitHub
Gen Code — multi-search, multiple AI providers, personas, skills & extensions, self-evolving agents

Open by design

Every layer is swappable at runtime. No lock-in.

🔌 Any LLM provider

Seven providers, from Anthropic to Z.ai. Switch with /model.

🔍 Pluggable search

Exa, Tavily, Brave, or Serper. Switch with /search.

🧩 Skills & extensions

Skills, plugins, MCP servers, and subagents — Claude Code compatible, run unmodified.

🎭 Personas

Markdown-defined agent identities, scoped per user or project, plus lifecycle hooks.

⚡ Native performance

A single Go binary. 12 MB, ~0.01s startup, no runtime.

💾 Session persistence

Sessions auto-save, resume, and fork, with automatic context compaction.

✦ Self-evolving  Coming soon

A background reviewer distills each session into durable memory and reusable skills — so the agent improves with use, no manual tuning.

Lv.1 Curious → Lv.10 Better → Lv.50 Stronger → Lv.∞ Legendary

Faster & leaner than Claude Code

Same model (claude-sonnet-4-6) on Apple Silicon, comparable feature set.

MetricGen CodeClaude CodeAdvantage
Download size12 MB63 MB + Node 112 MB5× smaller
Disk footprint38 MB175 MB4.6× smaller
Startup time~0.01s~0.20s20× faster
Startup memory~32 MB~189 MB5.8× less
Simple task~2.4s / 39 MB~10.4s / 286 MB4.3× faster
Tool-use task~3.3s / 39 MB~26.0s / 285 MB7.9× faster

Go native compilation vs Node.js V8/JIT/GC overhead. Reproduce the numbers →

Try it in 30 seconds

Open source under Apache 2.0. Star the repo if it's useful — it genuinely helps.

★ Star on GitHub Get started