Skip to content
| Marketplace
Sign in
Visual Studio Code>Programming Languages>Cinnacode - AI Code AssistantNew to Visual Studio Code? Get it now.
Cinnacode - AI Code Assistant

Cinnacode - AI Code Assistant

cinnacode

|
8 installs
| (0) | Free
AI-powered coding assistant with chat, completions, and slash commands. Pay-as-you-go with prepaid credits. No subscriptions.
Installation
Launch VS Code Quick Open (Ctrl+P), paste the following command, and press enter.
Copied to clipboard
More Info

Cinnacode — AI Coding Assistant for VS Code

Pay-as-you-go AI coding help. No subscription. Chat, inline fixes, and one-keystroke edits, powered by Claude — right inside your editor.

Get 200 free credits → · Docs · Pricing


Why Cinnacode?

Most AI coding assistants charge a flat monthly subscription — whether you use them heavily or barely touch them that month. Cinnacode works differently: buy credits once, spend them at your own pace. No recurring charge for the weeks you don't code.

  • 🎁 200 free credits on signup — no card required, whether you sign up as a guest or with GitHub
  • 💳 28 credits per chat message, 6 per inline completion
  • 📅 Credits valid for 9 months — no rush, no expiry pressure

Features

💬 Sidebar Chat

Ask questions, request fixes, or get explanations — right in a native VS Code sidebar panel. No new app, no context switch.

🤖 Agent Mode

Hand off a multi-step task — "add error handling to every route in this file," "write tests for this module" — and Cinnacode plans, writes files, and runs commands to get it done, showing you a diff and asking for approval before every single change. Nothing executes without your explicit sign-off. Start it via AI: Start Agent Task from the Command Palette.

⌨️ Cmd+K Quick Edit

Select any code, press Cmd+K (Mac) or Ctrl+K (Windows/Linux), describe the change in plain English, and review a diff before it touches your file.

⚡ Local Mode — free, offline, no credits used

Run a small local model on your own machine (via Ollama, free and open-source) to answer simple, trivial questions without spending credits or needing an internet connection. Available in both Sidebar Chat and Agent Mode's reply box — pick "Local" from the dropdown once it's set up. Cinnacode always checks first whether a question actually needs Claude's real reasoning; anything requiring genuine analysis, code generation, or multi-step logic still goes to Claude automatically, even with Local selected. Set it up anytime via AI: Enable Local Mode from the Command Palette — Cinnacode never installs anything or downloads a model without your explicit confirmation first.

Setting up Local Mode — click to expand

Cinnacode doesn't install anything for you automatically. Two things need to be on your machine first: Ollama (the runtime) and one small model (~1GB).

1. Install Ollama

  • macOS: download from ollama.com/download, or via Homebrew: brew install ollama
  • Windows: download the installer from ollama.com/download
  • Linux: curl -fsSL https://ollama.com/install.sh | sh

2. Confirm it's running

ollama --version

On macOS with Homebrew, if it's not running yet: brew services start ollama. On Windows/the direct macOS installer, launching the app once is enough — it runs in the background afterward.

3. Pull the model Cinnacode uses

ollama pull qwen2.5:1.5b-instruct-q4_K_M

This is a one-time download, roughly 1GB.

4. Enable Local Mode in Cinnacode

Open the Command Palette (Cmd+Shift+P / Ctrl+Shift+P) → run AI: Enable Local Mode. If Ollama and the model are already installed (steps 1–3 above), this finishes immediately. If you skip straight to this step without installing anything first, Cinnacode will detect what's missing and walk you through it.

5. Select it

In Sidebar Chat or Agent Mode's reply box, switch the provider dropdown from Claude to ⚡ Local.

🔍 Diff & Apply, Always

Every suggestion — from chat or Cmd+K — shows a real diff before anything changes. Apply stays disabled until you've actually reviewed it, and one more confirmation dialog stands between you and any file write. Nothing happens silently.

📎 @ File Mentions

Pull other files into a conversation by typing @ and a filename. Cinnacode reads them fresh, so it understands more than just the file you have open.

📄 Document Upload

Attach a requirements doc, spec, or any text-based file via the + button in the chat panel. Cinnacode can compare it against your actual code — pointing out what's implemented, what's missing, and proposing the code for the rest.

🗺️ Project-Wide Context

Ask "where is the discount calculation logic?" without knowing which file it's in — Cinnacode scans your whole project's structure to find and name the relevant file, even ones you've never opened or mentioned.

🎚️ Model Selection

Pick ⚡ Fast, ⚖️ Balanced, or 🧠 Most Capable right from the chat panel — trade off speed for depth depending on what you're working on. Your choice is remembered across sessions.

⚡ Slash Commands

Command What it does
/fix Returns the complete corrected file, plus what was fixed
/explain Explains what the code does, in plain terms
/improve Suggests readability, performance, and best-practice improvements
/test Writes unit tests covering main functionality and edge cases; auto-detects your project's test framework
/edge-cases Digs specifically into boundary values, empty/null inputs, and adversarial scenarios
/coverage Compares your code against its existing test file and flags what's not covered
/docs Writes documentation or comments for your code

⚠️ Important — Please Read

Cinnacode uses AI models to generate code, suggestions, and analysis. Like all AI coding tools, it can produce incorrect, incomplete, or misleading output — including references to packages or APIs that don't exist. Always review AI-generated code before using it in production.

Agent Mode can write files and run commands on your machine. Every action requires your explicit approval — but approving an action means you've reviewed it, not that Cinnacode has verified it's safe or correct. Read each proposed action carefully before approving.

Local Mode answers come from a small model running on your own machine, not Claude — it's deliberately used only for simple questions, but the same review principle applies: treat a local answer with the same scrutiny as any other AI-generated output.

Cinnacode is a tool to support your work, not a replacement for your own judgment or a qualified professional's review — particularly for code, documents, or analysis involving compliance, legal, financial, or safety-critical decisions.

Pricing

Plan Credits Price
Free 200 ₹0 — no card required
Starter 1,000 ₹199
Popular 4,000 ₹499
Power 10,000 ₹999

Full pricing and details: cinnacode.in/#pricing

Getting Started

  1. Install Cinnacode (you're already here!)
  2. On first launch, choose Login with GitHub or Continue as Guest
  3. Open the Cinnacode icon in the Activity Bar
  4. Start chatting, or select code and press Cmd+K

Full setup guide: cinnacode.in/resources.html

Commands

Access these from the Command Palette (Cmd+Shift+P / Ctrl+Shift+P):

  • AI: Start Agent Task
  • AI: Enable Local Mode
  • AI: Login with GitHub
  • AI: Logout
  • AI: Check Login Status
  • AI: Buy Credits
  • AI: Quick Edit Selection (bound to Cmd+K / Ctrl+K when text is selected)
  • AI: Send Feedback

Privacy & Security

  • We never see or store your payment card details — all payments are processed directly by Razorpay
  • Your code is sent to Anthropic's Claude API only to generate a response, never used to train any model
  • Forgot your password? Reset it anytime at cinnacode.in/login.html
  • Full details: Privacy Policy · Terms

Support

  • 📚 Docs & FAQ
  • 💬 Run AI: Send Feedback from the Command Palette
  • ✉️ Contact us

Cinnacode is built and maintained by an individual developer, not a large team — but every bit of feedback gets read.

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