Skip to content

What is Gee-Code?

Gee-Code is a general-purpose creation platform that runs in your terminal. You can use it to create code, documents, art, interfaces — really anything digital.

It works in two ways: you can operate it directly, typing natural language to build and create alongside the AI. Or you can deploy autonomous agents — called Gees — that can live, evolve, communicate with each other and with you, and accomplish incredible things independently.

Gee-Code isn’t locked to one AI provider. It supports a vast range of models:

  • Anthropic — Claude Opus, Sonnet, Haiku
  • OpenAI — GPT-4o, o1, o3
  • Google — Gemini Pro, Flash
  • xAI — Grok
  • Groq — Fast inference for Llama, Mistral
  • Cerebras — Ultra-fast inference
  • Ollama — Run local models (DeepSeek, Llama, Mistral)
  • OpenRouter — Access hundreds of models through one API
  • Custom — Any OpenAI-compatible API endpoint

Switch models mid-conversation with /model. Run different models for different agents. Use local models for privacy-sensitive work and cloud models for maximum capability.

Gee-Code ships with 100+ purpose-built tools that go far beyond code completion:

  • File operations — read, write, edit, search, glob across your entire filesystem
  • Code execution — run shell commands with timeout control and background support
  • Web browsing — fetch pages, search the web, automate browsers
  • Media generation — create images and video with FAL AI models
  • Git workflows — safe commit, branch, push with co-author attribution
  • Browser debugging — connect to Chrome DevTools, inspect DOM, read console, take screenshots
  • Notebook editing — work with Jupyter notebooks
  • Communication — send SMS, email, or voice notifications

Every tool has safety guardrails. Dangerous operations require explicit approval.

Gee-Code enhances everything you already do:

  • Persistent memory — it remembers your preferences, project context, and coding style across sessions
  • Intelligent context — it understands your codebase, tracks what files matter, and manages token budgets automatically
  • Task planning — break complex work into parallel execution graphs with dependency tracking
  • Session continuity — pick up exactly where you left off, even after days away
  • Deep codebase understanding — index and explore projects with /init and /explore

This is where Gee-Code goes beyond any other tool. You can deploy autonomous AI agents — Gees — each with:

  • A mission — what they exist to accomplish
  • Guardrails — what they’re allowed and not allowed to do
  • Objectives — specific work items to complete
  • An evolving identity — they develop a sense of self over time
  • Inter-mode communication — they can message each other and collaborate

Gees can run in the background as daemons, activating on heartbeat schedules or in response to triggers. They can work solo, in teams, or as part of larger endeavors.

You can also run flights (multiple agents in parallel on one task) and swarms (multi-cluster orchestrations with dependencies) for complex, high-throughput work.

  • Individual developers who want an AI that truly understands their codebase and works alongside them
  • Teams building AI-augmented development workflows
  • Power users who want full control over AI behavior — models, agents, guardrails, and autonomous operation
  • Creators who want to build anything digital, not just code