The two AI layers in KilatKoding
KilatKoding includes two different AI layers:
- AI features for your product,
- AI coding assistant configuration for your development workflow.
AI features for your product
Available routes
| Route | Purpose |
|---|
/api/ai/chat | Streaming chat |
/api/ai/generate | One-shot text generation |
Core files
| File | Purpose |
|---|
lib/ai/provider.ts | Selects the provider and model |
lib/ai/middleware.ts | Handles auth check, provider check, and usage limits |
lib/ai/usage.ts | Tracks token usage and plan limits |
hooks/use-ai-chat.ts | Client hook for chat UI |
Supported providers
| Provider | Required env | Default model as of March 19, 2026 |
|---|
| OpenAI | OPENAI_API_KEY | gpt-4o |
| Anthropic | ANTHROPIC_API_KEY | claude-sonnet-4-20250514 |
Token limits by plan
| Plan | Monthly token limit |
|---|
FREE | 0 |
BASIC | 10_000 |
PRO | 100_000 |
ULTIMATE | Unlimited |
AI request rate limits
| Plan | Request limit |
|---|
FREE | 10 requests per 5 minutes |
BASIC | 20 requests per 5 minutes |
PRO | 40 requests per 5 minutes |
ULTIMATE | 80 requests per 5 minutes |
How to disable AI
If your product does not need AI yet:
NEXT_PUBLIC_ENABLE_AI=false
| File | Tool |
|---|
AGENTS.md | General agent guidance and repository workflow |
CLAUDE.md | Claude Code |
.github/copilot-instructions.md | GitHub Copilot |
.cursorrules | Cursor |
.windsurfrules | Windsurf |
When you should touch this area
- you want to enable AI features for end users,
- you want to change the default provider,
- you want to adjust token limits by plan,
- you want to update the instructions used by your team’s AI coding assistants.
When you can ignore this area
- you do not need AI features in the product yet,
- you are only changing branding, content, or marketing pages.
If you are focused on shipping the MVP quickly, AI can stay disabled at first. KilatKoding is still valuable even without turning the AI layer on.
