⚡ AI Terminal v6 — VS Code Extension
Chat-style AI terminal assistant — all in one VS Code window. Paste any API key, type what you want, see the command, click Run.
What's new in v6
|
v5 |
v6 |
| UI |
Input box popups |
Built-in chat panel (opens beside your editor) |
| Window |
Required second window for testing |
One window — no F5 / Extension Host needed after install |
| JSON parse errors |
Could fail on Groq/Llama responses |
Fixed — handles single-quoted JSON, extra text, markdown fences |
| Run commands |
Auto-ran immediately |
Click ▶ Run button in chat — or 📋 Copy |
How it works
You type: "create a folder named vaibhav on the Desktop"
AI replies: New-Item -ItemType Directory -Path "$env:USERPROFILE\Desktop\vaibhav"
[▶ Run in Terminal] [📋 Copy]
Click Run → command executes in the VS Code terminal below
Supported API Keys (auto-detected)
Setup
Prerequisites
Step 1 — Get a free API key
Step 2 — Open the project
cd ai-terminal-v5 # or wherever you extracted the zip
code .
Step 3 — Install & build
npm install
npm run watch # leave this running
Wait for: [watch] build finished, watching for changes...
Step 4 — Launch (one-time for development)
Press F5 → Extension Development Host window opens.
After packaging as .vsix (see below), you never need F5 again — it works directly in your normal VS Code.
Step 5 — Open the chat panel
- Press
Ctrl+Shift+; (Mac: Cmd+Shift+;)
- OR click ⊙ AI Terminal in the status bar
- OR open Command Palette →
AI Terminal: Open
The chat panel opens on the right side of your editor.
Step 6 — Paste your API key
The panel shows a key input on first launch. Paste your key → click Connect →
Provider is detected automatically and confirmed.
Usage
| Action |
How |
| Ask AI |
Type in the chat box → Enter or click ➤ |
| Run command |
Click ▶ Run in Terminal in the chat bubble |
| Copy command |
Click 📋 Copy |
| Change API key |
Click 🔑 in the panel header, or Command Palette → AI Terminal: Set API Key |
| Clear chat |
Click 🗑 in the panel header |
| Override model |
Settings (Ctrl+,) → search aiTerminal.model |
Install permanently (skip F5 forever)
npm install -g @vscode/vsce
vsce package
code --install-extension ai-terminal-6.0.0.vsix
Now it works in your normal VS Code like any other extension.
Troubleshooting
Panel doesn't open
→ Make sure npm run watch shows "build finished" before pressing F5.
"Could not parse AI response" (old error — fixed in v6)
→ Update to v6. The new parser handles single-quoted JSON from Groq/Llama models.
API error 401
→ Key is invalid or expired. Click 🔑 and paste a fresh key.
"Unknown API key format"
→ The key prefix wasn't recognized. You'll be asked if you want to use it anyway (works for OpenAI-compatible APIs).
Changes not reflected
→ npm run watch auto-recompiles. Reload the Extension Host: Ctrl+Shift+P → Developer: Reload Window.
Settings
| Setting |
Description |
Default |
aiTerminal.model |
Override model name (blank = smart default) |
(blank) |
aiTerminal.shell |
Shell: auto, bash, zsh, fish, powershell, cmd |
auto |
Project structure
ai-terminal-v5/
├── src/
│ ├── extension.ts ← Entry point, WebviewPanel, message routing
│ ├── panelContent.ts ← Full chat UI (HTML/CSS/JS)
│ ├── llmProvider.ts ← Universal AI caller + robust JSON parser
│ └── terminalExecutor.ts← Shell/CWD detection
├── out/ ← Compiled (auto-generated)
├── package.json
└── tsconfig.json
License
MIT
#� �A�I�-�t�e�r�m�i�n�a�l�
�
�