Kolega Code

Kolega Code is a local-first AI coding assistant for developers working in the terminal or an editor. Its Gigacode engine lets a model write a Python workflow to coordinate sub-agents for repository reviews, migrations and tasks with independent workstreams. You can inspect and rerun those workflows, including parallel work and separate verification steps.

The agent runs locally. It supports local Llama models as well as cloud providers such as Anthropic, OpenAI, Google, Mistral AI and DeepSeek. You can assign different models to planning, coding or investigation, rather than use one model for every task. Sessions, settings, permissions and credentials stay on your machine; choosing a cloud model sends model requests to that provider. The project uses Business Source License 1.1.

For everyday changes, a read-only Plan mode investigates code and produces a task list, while Build mode edits files and runs shell commands under project permission controls. The terminal interface shows sub-agent activity and code diffs. Interrupted workflows can resume using saved results for calls whose inputs haven't changed, even if the workflow has been reordered.

Zed and VS Code can host the agent through the Agent Client Protocol, with approval prompts and session restoration. MCP support connects external tools, and browser agents can work with interactive websites through Playwright. Web search supports DuckDuckGo and self-hosted SearXNG, among other backends. One-shot requests with JSON output also support scripting and CI.