
OpenClaw AI Assistant for VS Code
概览
AI coding assistant for VS Code, powered by OpenClaw. - 💬 - Sidebar and popup panel chat interface - 🔄 - Visual preview and apply code changes - 🎯 - Per-session model override, multi-window independent - 🔌 - Live WebSocket connection indicator - 📨 - Send messages while AI is responding, auto-queued - 🔧 - Real-time tool invocation feedback via auto-refresh polling - 💡 - Smart error classification with actionable suggestions - 🎯 - Auto-detect and use project skills - 📎 - Attach code files and images to your messages - 🖼️ - Paste images from clipboard, auto-sent as multimodal vision attachments - 📁 - Drag files from Explorer, file tree, or editor tabs - 🔄 - Real-time model catalog via Gateway RPC - 🔄 - Up to 5 parallel chat sessions with independent history - 🌍 - Full i18n for UI and AI responses (zh-CN, en, ja, ko) - 🪟 - Enhanced Windows/WSL compatibility with auto-fallback - — Pasted/dropped images are sent as base64 attachments via Gateway's chat.
README
以下为原文 README 节选,完整内容请查看原文。
OpenClaw VS Code Extension
AI coding assistant for VS Code, powered by OpenClaw.
Features
- 💬 Chat with AI - Sidebar and popup panel chat interface
- 🔄 Diff Preview & Apply - Visual preview and apply code changes
- 🎯 Session-level Model Switching - Per-session model override, multi-window independent
- 🔌 Real-time Connection Status - Live WebSocket connection indicator
- 📨 Message Queue - Send messages while AI is responding, auto-queued
- 🔧 Tool Call Display - Real-time tool invocation feedback via auto-refresh polling
- 💡 Friendly Error Messages - Smart error classification with actionable suggestions
- 🎯 Skills & Workflows - Auto-detect and use project skills
- 📎 File & Image Attachments - Attach code files and images to your messages
- 🖼️ Image Paste & Vision - Paste images from clipboard, auto-sent as multimodal vision attachments
- 📁 Drag & Drop Files - Drag files from Explorer, file tree, or editor tabs
- 🔄 Dynamic Model List - Real-time model catalog via Gateway RPC
- 🔄 Multi-window Support - Up to 5 parallel chat sessions with independent history
- 🌍 Multi-language - Full i18n for UI and AI responses (zh-CN, en, ja, ko)
- 🪟 Windows & WSL Support - Enhanced Windows/WSL compatibility with auto-fallback
What’s New in v0.2.24
🖼️ Image Multimodal Vision
- Native vision transport — Pasted/dropped images are sent as base64 attachments via Gateway’s
chat.send, injected into the LLM’s multimodal image content block - Auto disk image conversion — Image files (
.png/.jpg/.jpeg/.gif/.webp/.bmp/.svg) referenced via attachment button or drag-drop are auto-read from disk and converted to base64 attachments - Text deduplication — Image paths already converted to attachments are removed from message text
🔄 Dynamic Model List via RPC
- RPC-first — Model list fetched via Gateway RPC
models.list+sessions.list, reflecting runtime model catalog - Config file fallback — Falls back to local config files if RPC fails
- Fuzzy matching — Current model always appears in the list
Installation
From GitHub Releases (Recommended)
- Download the latest
.vsixfrom Releases - Open VS Code
- Press
Ctrl+Shift+P→ “Extensions: Install from VSIX…” - Select the downloaded file
From Open VSX
- Open VS Code / VSCodium
- Go to Extensions (
Ctrl+Shift+X) - Search for “OpenClaw”
- Click Install
Requirements
- OpenClaw must be installed and running
- Gateway should be accessible at
http://localhost:18789(default)
Windows Users
If you encounter “Cannot find openclaw” error:
-
Find your OpenClaw path:
where openclaw -
Configure in VS Code Settings (
Ctrl+,):- Search for “OpenClaw: Openclaw Path”
- Enter the path, e.g.:
推荐工具
换一个关键词,或者移除筛选条件。
安装
npx skillfish add shenyingjun5/openclaw-vscode