Chat builder
Get the most out of OpenV2's conversational AI builder — prompts, modes, and tools.
The chat builder is the primary way you create and refine apps on OpenV2. Streaming AI turns your messages into real React Native code, schema changes, and configuration.
How chat works
- You send a message (text, voice, or with attachments)
- OpenV2 selects relevant files for context
- The AI streams a response with structured actions (create file, edit file, PocketBase schema, etc.)
- Actions apply to your project workspace
- Preview updates automatically
Each turn consumes build credits based on token usage.
Writing effective prompts
Include:
- Screens and what each shows
- User actions (tap, swipe, submit)
- Data — local, mocked, or V2 Cloud
- Visual style or a design reference
See Your first app for examples.
Chat modes & tools
| Tool | Description |
|---|---|
| Plan mode | Approve a plan before code runs |
| Visual editor | Click-to-edit styles in preview |
| Sketch | Wireframe before building |
| Prompt enhancer | Improves your message before sending |
| Voice input | Dictate prompts via microphone |
| Attachments | Include files for context |
| Device integrations | Native feature shortcuts |
| Payments panel | Connect RevenueCat and scaffold IAP |
| Version restore | Roll back to a prior snapshot |
AI models
The platform IDE uses Kimi K2.6 as the main generation model (configurable server-side). Helper tasks (summaries, file selection, prompt enhancement) use faster models like Gemini Flash Lite.
App icon generation uses FLUX.2 Klein 4B.
Skills
OpenV2 loads specialized skills automatically when relevant — layout rules, theming, monetization, debugging, and more.
Project knowledge
Add persistent rules in V2 Cloud Knowledge for app-specific instructions.
Collaborators & comments
Invite teammates as editors or viewers. Use header comments for project discussion.
Credits
Monitor usage in Settings → Usage. See Plans & credits and tips to save credits.