Radzor is a component registry with machine-readable manifests. Copy source code into your project with one command — your AI reads the manifest and integrates it for you.
$ npx radzor@latest add audio-capture ✓ @radzor/audio-capture@0.1.0 components/radzor/audio-capture/src/index.ts components/radzor/audio-capture/radzor.manifest.json ✓ Done! 2 files added.
{
"name": "@radzor/audio-capture",
"inputs": [{ "name": "sampleRate" }],
"actions": [{ "name": "startRecording" }],
"events": [{ "name": "onSpeechStart" }]
}Three steps. No config, no lock-in, no black-box packages.
npx radzor@latest add audio-capture copies the source code directly into your project. Like shadcn/ui — the code is yours.
Each component ships with a radzor.manifest.json describing its inputs, outputs, actions, and events.
Your AI reads the manifest and generates the integration code. No documentation hunting, no guessing APIs.
Standardized building blocks ready for LLM-driven development.
Browser-based audio recording with configurable sample rate, format, and real-time streaming.
Multi-provider OAuth 2.0 authentication with session management and token refresh.
Complete Stripe payment integration with subscriptions, webhooks, and refund handling.
WebSocket-based real-time messaging with rooms, typing indicators, and message history.
Multi-provider LLM text generation with streaming, multi-turn conversations, and tool calling.
Transactional email delivery via SMTP or API providers with templates and attachments.
Local and S3-compatible file uploads with validation, presigned URLs, and metadata tracking.
Token bucket and sliding window rate limiting with configurable quotas and automatic cleanup.
Each component declares typed inputs and outputs. Your LLM reads the manifest and automatically chains them into complete features.
Capture voice → transcribe → generate response → speak it back
Scrape content → store embeddings → query with LLM
Create checkout → receive webhook → send confirmation
Powered by composability in each manifest
Documentation is for humans. Manifests are for LLMs. A radzor.manifest.json describes everything an AI needs to integrate a component — inputs, outputs, actions, events, and composability rules.
Tell us which component costs you the most time, and we'll build it — with a full manifest, LLM docs, and production-ready code.
Join the community, post in #components-requests, and describe what you need. We prioritize based on real demand.
Browse the registry, add a component, and let your LLM handle the integration. Or build your own and share it with the community.