Meet your new workspace.

Install & get started

Download the HuyIDE build for your operating system. Open the app, complete the initial setup, then open your project folder to begin.

Connect your AI

Open AI Provider Hub, choose a provider, and follow its connection instructions. Your provider may require an API key, an account sign-in, or a running local service such as Ollama.

Work with agents

Open chat in your workspace, describe the desired outcome, and include relevant files or context. For larger tasks, ask for a plan before implementation. Follow the commands and changes made during the session.

Check in the browser

Run your web project, open its address in the integrated browser, and ask AI to check specific interactions. Describe the action and expected result so the agent has a clear target.

Review changes

Inspect the changed files and diff. Run the checks appropriate to your project, compare the behavior with your request, then commit using Git.

MCP & extensions

Connect MCP servers to add tools for AI. Install extensions from Open VSX for the languages and workflows you need.

NEW / CMD + CMD

Quick capture. Straight into chat.

Press left Cmd + right Cmd together on macOS to capture an application window and attach it to HuyIDE chat. Add a prompt to help AI understand a UI, investigate a bug, or assist with code changes.

1. Run Enable Global Screenshot Shortcut in the Command Palette. Grant HuyIDE Accessibility and Screen Recording permissions in macOS settings; enable the shortcut again or restart the app if requested.
2. Open the window to capture, such as Safari, then press left Cmd + right Cmd together. When you return to HuyIDE, the last external window is the capture source.
3. Review the attachment in chat, add your prompt, then send it to an image-capable model.

01 / ACCOUNTS & APIs

Your AI. Connected in one place.

Sign in to providers with your existing accounts, add API keys, or use local models. Manage multiple OpenAI-compatible and Anthropic-compatible endpoints inside HuyIDE.

1. Open AI Configuration → API Keys and select Connect for your provider.
2. For Custom, choose Add endpoint. Enter a name, URL, API key and the matching API type.
3. Select the endpoint’s models, save, then use them in chat or Model Aliases.

02 / HUYCOMPOSE · PRO

A model hits its limit. Your next choice is ready.

Put models in priority order. When a model fails, runs out of quota or hits a rate limit, HuyCompose tries the next model in your pool. Text and vision have separate pools.

1. Open AI Configuration → HuyCompose. Add models to the Text model or Vision model pool.
2. Use the up/down controls to set priority. Enable streaming if your provider requires it.
3. Select HuyCompose in chat or for the alias your agent uses.

03 / MODELS FOR EACH AGENT

The right model for each agent.

Assign a model or alias to each agent. Change an alias once to update the agents sharing it, or point it at HuyCompose to use your fallback pool.

1. Open Model Aliases, select an alias such as default/code and choose its model.
2. Open Agents, select an agent and set Language Model under LLM Requirements.
3. Check Evaluates to and Agents using this Alias to confirm your setup.

04 / SUPERPOWERS & SKILLS

Give your agents a way of working.

Skills provide reusable instructions. The Superpowers integration maps reading, editing, terminal, delegation and web tools to HuyIDE’s tools.

1. Open AI Configuration → Skills to browse skills and Curated Skill Plugins.
2. Install Superpowers from the list if needed, then inspect the Compatibility table.
3. Open a skill to read its instructions and use the relevant skill for your task.