FAQ
The questions we actually get asked.
Do you train on my code?+
No. Your code is never used for training. Prompts and completions are not stored — see Privacy.
Can I use my own model?+
Yes. CodexaX has 11 built-in providers (Anthropic, OpenAI, Gemini, xAI, OpenRouter, Ollama, Azure, OpenAI Compatible…). Paste your key once, switch any time.
Is CodexaX a fork of VS Code?+
The editor is a fork. Your existing VS Code extensions, themes, and settings work unchanged. The new parts (chat, agents, MCP, skills, rules) are bundled.
Does it work offline?+
The editor itself is fully offline. AI features need a network unless you point them at a local Ollama instance.
Can I self-host?+
Yes — both the editor and backend are open source. See Self-hosting.
What about my Cursor / Continue setup?+
Cursor's
.cursorrules imports automatically into .codexax/rules.md. Continue's MCP config drops in unchanged.Can I cancel anytime?+
Yes. Cancel from Console → Billing; you keep access until the end of the period. 14-day refund on monthly Pro.
Do you support Windows on ARM?+
Yes. Native ARM64 builds for Windows 11 and Linux. macOS is universal (Apple Silicon + Intel).
Is there a free tier?+
Yes — Hobby is 200 chat messages / month with all editor features.
How is this different from GitHub Copilot?+
You're not locked to GitHub auth or one model vendor. Chat / agents / skills are first-class. And you can self-host.