Code Buddy
A local AI coding agent for Linux, macOS and Windows that edits repositories, runs verification commands and supports Ollama or cloud models.

Code Buddy is a terminal AI coding agent for developers who want it to read a repository, make changes and run commands on their machine. It supports local models through Ollama as well as cloud providers.
- Verification loop: The agent plans and edits code, then checks its work with a verification command such as a test suite and continues until that command passes.
- Isolated edits: An optional workspace lets it validate proposed writes before they reach your working files.
- Parallel tasks: It can divide independent work among subagents and search across multiple repositories with read-only access.
- Persistent context: Project and user memory retain lessons between sessions, while session history supports returning to earlier work.
- Optional learning: It can propose tools, skills and execution strategies, evaluate them against a snapshot, and reject changes that show no improvement or cause regressions; this changes its behavior rather than model weights.
The CLI runs on Linux, macOS and Windows. Ollama inference runs locally without an API key; cloud model requests go to external providers. Self-hosted peers can call each other's models and share restricted, read-only tools. A Claude Code plugin provides a read-only MCP connection, and a Codex skill lets that assistant work with Code Buddy.