Sentinel Coder

Multi-provider AI coding assistant for VS Code — 12 providers, 50+ models, local + cloud. A free, open-source alternative to GitHub Copilot.

Built by QubitPage Research.

Features

• 12 AI Providers — Ollama (local), Groq (free), Google Gemini (free tier), HuggingFace (free), OpenAI, Anthropic, DeepSeek, Mistral, Together AI, OpenRouter, Vultr, and custom OpenAI-compatible endpoints
• 50+ Models — GPT-4o, Claude Sonnet 4, Gemini 2.5 Pro, Llama 3.3 70B, Qwen3, DeepSeek V3, and many more
• Auto Model Router — Automatically picks the best model for each task (coding, reasoning, speed, agentic)
• Agent Mode — 18 built-in tools: create files, read code, run terminals, search workspace, serve HTML, open browser
• Thinking Display — See the model's reasoning process in a collapsible panel (for thinking models like o3, DeepSeek R1, Qwen3)
• Sub-Agent Delegation — Delegate sub-tasks to specialized models (e.g. use a fast model for search, a reasoning model for analysis)
• Three Modes — Agent (autonomous tools), Ask (Q&A), Plan (analysis without execution)
• Rich Model Catalog — See pricing, context windows, and feature badges (Tools, Thinking, Vision, Streaming) for every model
• Zero Telemetry — All API keys stored locally via encrypted SecretStorage. No data leaves your machine except to the providers you choose.

Quick Start

1. Install from the VS Code Marketplace
2. Open the Sentinel Coder sidebar (click the S icon in the activity bar)
3. Pick a provider — Click the ⚙ gear icon → Providers tab → Enable a provider and enter your API key
4. Select a model from the dropdown
5. Start chatting!

Free Providers (no credit card needed)

Paid Providers

Modes

Agent Mode

The AI can autonomously use tools to complete tasks:

• File Tools — createFile, readFile, listDirectory, searchFiles, searchText
• Editor Tools — getActiveFile, getSelection, getDiagnostics, openFile
• Terminal Tools — runCommand
• Web Tools — serveFile, openBrowser, fetchUrl
• Analysis Tools — delegateSubAgent (spawn sub-tasks on different models)

Ask Mode

Direct Q&A — the AI answers questions and writes code without using tools.

Plan Mode

The AI analyzes your request and creates a step-by-step plan without executing anything.

Configuration

All settings are under sentinelCoder.* in VS Code settings:

API Keys File Format (Optional)

For power users who want to bulk-import keys from a text file:

groq: gsk_your_groq_key_here
openrouter: sk-or-v1-your_openrouter_key
gemini: AIzaSyYour_google_key
vultr inference: YOUR_VULTR_KEY
huggingface: hf_your_hf_token

Set sentinelCoder.apiKeysFile to the file path. Keys entered via the Settings > Providers panel are stored securely in VS Code's encrypted SecretStorage each.

Privacy & Security

• Zero telemetry — No analytics, no tracking, no data collection
• API keys encrypted — Stored in VS Code SecretStorage (OS-level encryption), never in plaintext settings
• Local-first — Works fully offline with Ollama. Cloud providers are optional
• Open source — MIT License. Inspect every line of code

Requirements

• VS Code 1.93.0 or later
• For local models: Ollama installed and running
• For cloud models: An API key from any supported provider

Known Issues

• The Chat Participant (@sentinel in the native chat panel) requires VS Code 1.93+
• Some thinking models may have a delay before the first response token

Release Notes

3.0.0

• Multi-provider support (12 providers, 50+ models)
• Auto model router — picks the best model per task
• Sub-agent delegation for complex workflows
• Thinking display for reasoning models
• Rich model catalog with pricing and feature badges
• Secure API key storage via SecretStorage
• CoderQ personality mode

Made with ❤️ by QubitPage Research