Smart AI Routing Architecture

@eldrex/plugin-study-buddy includes an intelligent three-tier AI routing system called StudyBuddyAIRouter that guarantees fast, privacy-safe, and resilient code explanations.

---

The 3-Tier Priority Cascade

┌────────────────────────────────────────┐
│               Incoming Request         │
└───────────────────┬────────────────────┘
                    │
                    ▼
┌────────────────────────────────────────┐
│  Tier 1: IDE Agent Context             │ (Fastest, zero config required)
└───────────────────┬────────────────────┘
                    │ (If outside IDE)
                    ▼
┌────────────────────────────────────────┐
│  Tier 2: Local Ollama Model            │ (100% private, runs entirely on CPU/GPU)
└───────────────────┬────────────────────┘
                    │ (If Ollama unavailable)
                    ▼
┌────────────────────────────────────────┐
│  Tier 3: User Cloud AI API Keys        │ (OpenAI, Anthropic, Google Gemini)
└────────────────────────────────────────┘

---

1. Tier 1: IDE Agent Context

When running inside VS Code, Cursor, or Windsurf, Study Buddy connects to the active editor's native agent runtime:

• Latency: < 100ms
• Cost: Free (uses active IDE subscription/session)
• Setup: Zero configuration required

---

2. Tier 2: Local Ollama (100% Private)

If you are running in a standalone terminal without an IDE agent, Study Buddy connects to local Ollama models:

• Recommended Model: llama3.2:3b or qwen2.5-coder:7b
• Data Privacy: Your code never leaves your local machine.
• Cost: Free forever

To start Ollama:

ollama run llama3.2:3b

---

3. Tier 3: Cloud AI Providers

If you configure API keys with DevDiff (devdiff auth add), Study Buddy can leverage powerful cloud models:

• Supported Providers: OpenAI (gpt-4o-mini), Anthropic (claude-3-5-haiku), Google Gemini (gemini-1.5-flash).
• Security: API keys are stored encrypted in your local OS keychain, and secrets/credentials are automatically redacted before transmission.