VulcanChat — AI Coding Assistant for VSCode

Vulcan is a feature maximalist, UI minimalist, AI-powered coding assistant for Visual Studio Code with ralph loops, queued prompts, smart prompts, cost calculator, agent tools, designed to replace cline/copilot.

Features

| 6 AI Models | DeepSeek , Kimi, OpenAI, Claude , Gemini, Minimax | | Ralph Loop | Repeat prompts on interval with $ cost limit | | Queued Prompts | Execute a list of queued prompts, press they go button to run them all, right click or shift click the queued button (under loop) to see the list, if you have ralph loop enabled it will loop through the queued prompts. | | Token Cost Calculator | Real-time $USD per model with running totals | | Smart Prompt | Custom system instruction appended to every conversation | | Plan/Act Mode | Plan mode (analyze only) vs Act mode (edit files) | | Visual Diffs | Side-by-side diff cards with Accept/Reject/Copy | | Clickable File Changes | +/- per file list, click to open in native VSCode diff | | Conversation Restore | Hover on any user prompt → revert all accepted changes | | Chat History | Per-workspace persistent history with load/delete | | 14 Agent Tools | agent, runCommand, toolSearch, vscodeAPI, newWorkspace, memory, askQuestions, webFetch, readFile, writeFile, listFiles, openEditor, searchProject, executeTerminal | | Terminal | Execute shell commands from chat |

Installation

From VSIX

1. Download vulcanchat-1.0.0.vsix
2. In VSCode: Ctrl+Shift+X → ... → "Install from VSIX..."
3. Select the downloaded file

From Source

# No npm install needed! Copy to your extensions folder:
mkdir -p ~/.vscode/extensions/vulcan-ai.vulcan-1.0.0
cp -r * ~/.vscode/extensions/vulcan-ai.vulcan-1.0.0/
# Reload VSCode window (Ctrl+Shift+P -> "Developer: Reload Window")

Setup

1. Press Ctrl+Shift+P → "Vulcan: Set API Key" or open Settings (⚙) → API Keys to add multiple provider keys
2. Enter an API key for any supported provider (DeepSeek, Kimi, OpenAI, Claude, Gemini, or Minimax)
3. Select your model from the settings dropdown (20 models across 6 providers)
4. Open the chat: click the flame icon in the activity bar or press Ctrl+Shift+L

Commands

Command	Shortcut	Description
Vulcan: Open Chat	Ctrl+Shift+L	Open the chat sidebar
Vulcan: Explain This	Right-click	Explain selected code
Vulcan: Fix This	Right-click	Find and fix bugs
Vulcan: Edit Selection	Right-click	Apply an edit instruction
Vulcan: Set API Key	Palette	Store your API key securely

Settings Bar Controls

• Depth — Lines of surrounding code context sent to the AI around your cursor (3–50)
• Model — Choose from 20 models across 6 providers (DeepSeek · Kimi · OpenAI · Claude · Gemini · Minimax)
• Max Tokens — Output token limit (256–131,072)
• API Keys — Collapsible section to store keys per provider
• Smart Prompt — Custom system instruction appended to every conversation
• Tools — Toggles to enable/disable read files, edit files, terminal, web fetch, and run commands
• Plan/Act — Toggle between planning mode and edit mode

Architecture

vulcan/
├── package.json              Extension manifest
├── README.md                 This file
├── media/
│   ├── chat.css              VSCode-native theme styling
│   ├── chat.js               Frontend (streaming, diffs, settings)
│   └── vulcan-icon.png       Extension icon
└── src/
    ├── extension.js          Activation & commands
    ├── chatProvider.js       Webview lifecycle & messaging
    ├── api.js                DeepSeekClient & PromptBuilder
    ├── storage.js            Per-workspace chat history & diffs
    ├── tools.js              14 built-in agent tools
    ├── terminalProvider.js   Terminal execution
    └── types.js              Model pricing & endpoint routing

Requirements

• VSCode ^1.85.0
• An API key for any supported provider: DeepSeek, Kimi, OpenAI, Claude, Gemini, or Minimax (set via Vulcan: Set API Key or Settings → API Keys)

Development

No build step required — the extension is pure CommonJS JavaScript.

To test in Extension Development Host:

1. Open this folder in VSCode
2. Press F5

License

MIT