WebInfer

WebInfer Skill

Teach your AI coding assistant how to build with WebInfer. One command adds client-side AI knowledge to your project.

What is This?

WebInfer brings client-side AI to any website or web app — text generation, images, speech, audio, tools, and agents — all through a single SDK that connects to the user's own AI providers. No API keys in your code, no vendor lock-in.

A skill file is a markdown document that teaches AI coding assistants how to use a library. When you add the WebInfer skill to your project, tools like Claude Code, Cursor, GitHub Copilot, and OpenAI Codex automatically know how to write correct WebInfer code — the right imports, API patterns, and best practices.

How WebInfer works

Your frontend calls the WebInfer SDK → the user's browser extension, daemon, or cloud connection routes the request to their chosen provider (OpenAI, Anthropic, Google, Ollama, etc.) → your code never touches API keys.

Your App → WebInfer SDK → User's Provider

Installation

Add the WebInfer skill to your project with one command:

Skills CLI
Recommended
skills.sh ecosystem — works with all agents
AGENTS.md
Claude Code, Codex & other agents
.cursorrules
Cursor IDE
copilot-instructions.md
GitHub Copilot
CLAUDE.md
Claude Code

Install all targets at once

Working with multiple AI assistants? Install all skill file formats:

npx @webinfer/skill all

What the Skill Covers

The skill file teaches your AI assistant the full WebInfer API surface for building client-side AI features:

Text Generation

generateText, streamText — chat, completion, multi-turn conversations

Structured Output

generateObject — typed JSON output with schema validation

Image Generation

generateImage — create images from text prompts

Speech & Audio

generateSpeech, generateAudio, generateMusic — TTS, sound effects, music

Tools & Function Calling

tool() — define functions the LLM can call with typed parameters

Agents

createAgent — multi-step tool loops with configurable stop conditions

Plus: model routing, React hooks, OpenAI-compatible API, Vercel AI SDK integration, error handling, cancellation, and a clear set of rules for what NOT to do.

Quick Example

Once the skill is installed, your AI assistant can write code like this correctly:

AI-powered React component

Package Details

The @webinfer/skill package includes a CLI installer, the SKILL.md content, and a programmatic API.

CLI Reference
Programmatic API

Supported AI Assistants

Claude Code
CLAUDE.md / AGENTS.md

Anthropic's CLI coding assistant reads project context from CLAUDE.md and AGENTS.md automatically.

OpenAI Codex
AGENTS.md

OpenAI's Codex agent reads AGENTS.md for project-specific instructions and conventions.

Cursor
.cursorrules

Cursor IDE reads .cursorrules from your project root to understand project-specific conventions.

GitHub Copilot
.github/copilot-instructions.md

Copilot reads custom instructions from .github/copilot-instructions.md for project context.

Gemini CLI
AGENTS.md

Google's Gemini CLI reads AGENTS.md files for project-specific context.

Amazon Q / Amp
AGENTS.md

Amazon's coding assistants read AGENTS.md for project conventions.

Links