Skip to content
| Marketplace
Sign in
Visual Studio Code>Programming Languages>Cdoing AgentNew to Visual Studio Code? Get it now.
Cdoing Agent

Cdoing Agent

Muhammad Awais Shah

|
8 installs
| (0) | Free
AI-powered coding assistant with multi-model support — chat, instruct, and run commands right from VS Code
Installation
Launch VS Code Quick Open (Ctrl+P), paste the following command, and press enter.
Copied to clipboard
More Info

Cdoing Agent — AI Coding Assistant for VS Code

Open-source, multi-provider AI coding assistant that lives right inside VS Code. Chat with AI, edit code inline, get autocomplete suggestions, and run 20+ tools — all without leaving your editor.

Use your own API key with Anthropic (Claude), OpenAI (GPT), Google (Gemini), Ollama, or any OpenAI-compatible provider.

Cdoing Agent — VS Code Extension


Features

Chat Panel

  • Sidebar chat or editor panel (beside your code)
  • Real-time streaming responses
  • Multi-tab conversations with independent state
  • Conversation history — resume past chats
  • Markdown rendering with syntax-highlighted code blocks
  • Copy button on code blocks
  • Clickable file paths in responses

Inline Edit (Cmd+I)

  • Select code, press Cmd+I / Ctrl+I, type your instruction
  • See inline diff preview before accepting changes
  • Works without opening the chat panel

Inline Autocomplete

  • Ghost text suggestions as you type
  • Tab to accept, Esc to dismiss
  • Configurable model (use a faster model for autocomplete)

Code Actions

  • Right-click any selected code for quick actions:
    • Explain — get a detailed explanation
    • Refactor — improve code structure
    • Fix — find and fix issues
    • Add to Chat — send code as context
    • Send to Chat — ask about selected code

20+ Built-in Tools

The AI agent can autonomously:

  • Read, write, and edit files with multi-strategy matching
  • Search your codebase (glob patterns, regex, FTS5 indexed search)
  • Run shell commands (with background mode for servers)
  • Fetch web pages and search the web
  • Spawn sub-agents for parallel tasks
  • Track tasks with built-in todo system

Context Providers

Attach rich context using @ mentions in chat:

Trigger Description
@file <path> Include a specific file
@codebase <query> Search your entire codebase
@terminal Last terminal output
@tree Workspace file tree
@url <url> Fetch a web page
@git Branch, status, commits
@diff Current working changes
@open All open editor files
@problems Current file diagnostics
@clipboard Clipboard contents

Permission System

Control what the AI can do:

  • Ask — prompts before every tool that requires permission
  • Auto-Edit — auto-approves file edits, asks for shell commands
  • Auto — auto-approves everything (for trusted environments)

Multi-Provider Support

Provider Models Auth
Anthropic (default) Claude Sonnet 4.6, Opus 4.6, Haiku 4.5 API key or OAuth
OpenAI GPT-4o, GPT-4o mini, o3-mini API key
Google Gemini 2.0 Flash, 1.5 Pro, 1.5 Flash API key
Ollama LLaMA, Mistral, CodeLlama Not required
Custom Any OpenAI-compatible API Configurable

Getting Started

1. Install the extension

Search for "Cdoing Agent" in the VS Code Extensions panel, or install from the Marketplace.

2. Set your API key

Open VS Code Settings → search "Cdoing" → enter your API key. Or set an environment variable:

# Anthropic
export ANTHROPIC_API_KEY=sk-ant-...

# OpenAI
export OPENAI_API_KEY=sk-...

# Google
export GOOGLE_API_KEY=AI...

3. Start chatting

Click the Cdoing icon in the Activity Bar (left sidebar) and start typing.


Keyboard Shortcuts

Shortcut Action
Cmd+Shift+L Open chat as editor panel (beside code)
Cmd+I / Ctrl+I Inline edit selected code
Cmd+Shift+Enter Send selected code to chat

Settings

Setting Default Description
cdoing.provider anthropic AI provider
cdoing.model (auto) Model name (empty = provider default)
cdoing.apiKey (empty) API key (overrides env var)
cdoing.authMethod apiKey Auth method: apiKey or oauth
cdoing.customBaseURL (empty) Base URL for custom providers
cdoing.temperature 0 Model temperature (0 = deterministic)
cdoing.maxTokens 8096 Max tokens in response
cdoing.permissionMode ask Permission mode: ask, auto-edit, auto

Using Ollama or Local Models

  1. Set cdoing.provider to custom
  2. Set cdoing.customBaseURL to http://localhost:11434/v1
  3. Set cdoing.model to llama3 (or your model)
  4. Leave cdoing.apiKey empty

Slash Commands

Command Description
/help Show available commands
/clear Clear conversation
/new Start a new conversation
/compact Compress context
/plan Enter planning mode (read-only)
/history Browse past conversations
/effort Set response depth

Open Source

Cdoing Agent is fully open source. The extension is part of the cdoing-agent monorepo which also includes a CLI (@cdoing/cli) and core library (@cdoing/core).

Built with: TypeScript, React, esbuild, LangChain

Packages:

  • @cdoing/core — tools, permissions, sandbox, hooks
  • @cdoing/cli — terminal-based AI assistant

Feedback & Issues

  • GitHub Issues
  • GitHub Repository

License

MIT

  • Contact us
  • Jobs
  • Privacy
  • Manage cookies
  • Terms of use
  • Trademarks
© 2026 Microsoft