Open Source Alternatives
Open source alternatives to Claude Code
Open source alternatives to Claude Code, ranked by GitHub stars and freshness.
9 open-source alternatives in the index, ranked by GitHub stars and freshness.
stippi/code-assistant
Various
An LLM-powered, autonomous coding assistant. Also offers an MCP and ACP mode.
Best for: Developers seeking an autonomous coding assistant with flexible integration options
sim-xia/skill-cortex-server
Various
A third-party MCP server that enable all IDEs to access Claude Code Skills capabilities
Best for: Developers who want to use Claude Code Skills in IDEs that are not natively supported by the official tooling.
Aider
Paul Gauthier
Terminal-first AI pair programmer. Edits files in your repo, commits with sensible messages, runs your tests.
Best for: Engineers who live in the terminal and want a tight, scriptable AI loop
Cline
Cline
Open-source autonomous coding agent that lives inside VS Code. BYO model key, watch it work.
Best for: Engineers who want Cursor-class agent power without leaving stock VS Code
Cursor Background Agent
Anysphere
Cloud-based async coding agent that clones your repo, works on a branch, and ships a PR while you keep coding.
Best for: Teams who want to parallelize coding work across isolated cloud environments
GitHub Copilot
GitHub / Microsoft
The original AI pair programmer, now with Agent Mode, Workspaces, and multi-model selection.
Best for: Large orgs already on GitHub Enterprise
OpenAI Codex (CLI + cloud agent)
OpenAI
OpenAI's coding agent surface: terminal CLI for local work, cloud agent for parallel autonomous tasks.
Best for: Teams on the OpenAI Enterprise side of the fence
OpenHands
All Hands AI
Open-source autonomous coding agent platform. Spins up a sandboxed dev environment, ships PRs end to end.
Best for: Teams who want a real autonomous coding agent and can run their own infra
Windsurf
Codeium
Codeium's IDE-with-an-agent. Cascade flow combines deep context with autonomous multi-file work.
Best for: Engineers who want continuous agent context, not chat-and-snippet