diff --git a/.github/ISSUE_TEMPLATE/bug_report.yml b/.github/ISSUE_TEMPLATE/bug_report.yml
new file mode 100644
index 0000000..bf60a2e
--- /dev/null
+++ b/.github/ISSUE_TEMPLATE/bug_report.yml
@@ -0,0 +1,36 @@
+name: Bug report
+description: Report a problem with LearnPath
+title: "[Bug]: "
+labels: ["bug"]
+body:
+  - type: textarea
+    id: what-happened
+    attributes:
+      label: What happened?
+      description: A clear description of the bug, including any console/server errors.
+    validations:
+      required: true
+  - type: textarea
+    id: repro
+    attributes:
+      label: Steps to reproduce
+      placeholder: |
+        1. Set LLM_PROVIDER to ...
+        2. Open /quiz and ...
+        3. See error
+    validations:
+      required: true
+  - type: input
+    id: provider
+    attributes:
+      label: Provider and model
+      placeholder: openai / gpt-4o-mini
+    validations:
+      required: true
+  - type: input
+    id: node
+    attributes:
+      label: Node version
+      placeholder: "20.x"
+    validations:
+      required: true
diff --git a/.github/ISSUE_TEMPLATE/config.yml b/.github/ISSUE_TEMPLATE/config.yml
new file mode 100644
index 0000000..b2f872c
--- /dev/null
+++ b/.github/ISSUE_TEMPLATE/config.yml
@@ -0,0 +1,5 @@
+blank_issues_enabled: true
+contact_links:
+  - name: Question / discussion
+    url: https://github.com/aimanmalib/learnpath/discussions
+    about: Ask questions and discuss ideas before opening an issue.
diff --git a/.github/ISSUE_TEMPLATE/feature_request.yml b/.github/ISSUE_TEMPLATE/feature_request.yml
new file mode 100644
index 0000000..771b659
--- /dev/null
+++ b/.github/ISSUE_TEMPLATE/feature_request.yml
@@ -0,0 +1,31 @@
+name: Feature request
+description: Suggest an idea or improvement for LearnPath
+title: "[Feature]: "
+labels: ["enhancement"]
+body:
+  - type: textarea
+    id: problem
+    attributes:
+      label: Problem / motivation
+      description: What are you trying to do that LearnPath doesn't support today?
+    validations:
+      required: true
+  - type: textarea
+    id: proposal
+    attributes:
+      label: Proposed solution
+    validations:
+      required: true
+  - type: dropdown
+    id: area
+    attributes:
+      label: Area
+      options:
+        - New LLM provider preset
+        - New study feature
+        - SM-2 algorithm improvement
+        - UI / accessibility
+        - Documentation
+        - Other
+    validations:
+      required: true
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
new file mode 100644
index 0000000..0be1a04
--- /dev/null
+++ b/.github/workflows/ci.yml
@@ -0,0 +1,38 @@
+name: CI
+
+on:
+  push:
+    branches: [main]
+  pull_request:
+    branches: [main]
+
+jobs:
+  test:
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        node-version: ["18.x", "20.x"]
+
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Use Node.js ${{ matrix.node-version }}
+        uses: actions/setup-node@v4
+        with:
+          node-version: ${{ matrix.node-version }}
+          cache: npm
+
+      - name: Install dependencies
+        run: npm ci
+
+      - name: Type check
+        run: npx tsc --noEmit
+
+      - name: Run tests
+        run: npm test
+
+      - name: Build
+        run: npm run build
+        env:
+          MIMO_API_KEY: dummy-key-for-build
diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
new file mode 100644
index 0000000..f8d82d3
--- /dev/null
+++ b/CONTRIBUTING.md
@@ -0,0 +1,78 @@
+# Contributing to LearnPath
+
+Thanks for your interest in improving LearnPath. This project is a
+provider-agnostic adaptive learning platform (Next.js + TypeScript) that runs
+on any OpenAI-compatible LLM endpoint. Contributions of all sizes are welcome
+— docs fixes, new provider presets, UI improvements, and bug reports all help.
+
+## Getting started
+
+```bash
+git clone https://github.com/aimanmalib/learnpath.git
+cd learnpath
+npm install
+
+echo 'LLM_PROVIDER=openai' > .env.local
+echo 'OPENAI_API_KEY=*** .env.local
+
+npm run dev      # http://localhost:3000
+npm test         # 116 tests should pass
+```
+
+## Development workflow
+
+1. Fork the repo and create a feature branch: `git checkout -b feat/my-change`
+2. Make your change with tests.
+3. Run the full local gate before pushing:
+   ```bash
+   npx tsc --noEmit     # type check
+   npm test             # vitest
+   npm run build        # production build
+   ```
+4. Commit using [Conventional Commits](https://www.conventionalcommits.org/)
+   (`feat:`, `fix:`, `docs:`, `test:`, `refactor:`, `ci:`).
+5. Open a pull request against `main`. CI runs type check + tests + build on
+   Node 18/20 — keep it green.
+
+## Adding a new LLM provider
+
+LearnPath talks to any OpenAI-compatible `/chat/completions` endpoint, so most
+providers need only a preset entry. In `src/lib/mimo-client.ts`, add to
+`PROVIDER_PRESETS`:
+
+```ts
+myprovider: {
+  baseUrl: 'https://api.myprovider.com/v1',
+  authStyle: 'bearer',          // 'bearer' or 'api-key'
+  model: 'default-model-name',
+  envKey: 'MYPROVIDER_API_KEY',
+  envBase: 'MYPROVIDER_BASE_URL',
+},
+```
+
+Then add a test in `tests/lib/multi-backend.test.ts` mirroring the existing
+provider cases.
+
+## Good first issues
+
+- Add a provider preset (Together, Groq, DeepSeek, Mistral, ...)
+- Add a new study feature (essay grading, flashcard import/export)
+- Improve the SM-2 spaced repetition algorithm
+- Accessibility / responsive UI improvements
+
+## Code style
+
+- TypeScript with strict mode (see `tsconfig.json`)
+- Vitest for tests
+- Next.js App Router conventions
+- Keep components small and focused
+
+## Reporting bugs / requesting features
+
+Use the issue templates (bug report / feature request). Include repro steps,
+your provider/model, and your Node version for bugs.
+
+## License
+
+By contributing, you agree your contributions are licensed under the MIT
+License, the same as the project.
diff --git a/README.md b/README.md
index 4b1da63..ae42fdb 100644
--- a/README.md
+++ b/README.md
@@ -1,9 +1,17 @@
 # 🎓 LearnPath
 
-**Adaptive Learning Platform powered by Xiaomi MiMo V2.5 Pro**
+**Adaptive Learning Platform for any OpenAI-compatible LLM**
 
 > AI-driven personalized education — quizzes, flashcards, learning roadmaps, progress tracking, and concept mapping.
 
+[![Next.js 14](https://img.shields.io/badge/Next.js-14-black.svg)](https://nextjs.org/)
+[![TypeScript](https://img.shields.io/badge/TypeScript-5.4-blue.svg)](https://www.typescriptlang.org/)
+[![CI](https://github.com/aimanmalib/learnpath/actions/workflows/ci.yml/badge.svg)](https://github.com/aimanmalib/learnpath/actions/workflows/ci.yml)
+[![Tests: 116](https://img.shields.io/badge/tests-116-brightgreen.svg)](tests/)
+[![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](LICENSE)
+
+Works with **OpenAI, OpenRouter, Ollama, llama.cpp, Xiaomi MiMo**, or any endpoint that speaks the OpenAI `/chat/completions` protocol. Pick a provider with one env var — no code changes.
+
 ---
 
 ## Architecture
@@ -23,7 +31,7 @@
 │  └────────────────────┬───────────────────┘             │
 │                       │                                  │
 │  ┌────────────────────▼───────────────────┐             │
-│  │           Custom Hooks (useMiMo)        │             │
+│  │         Custom Hooks (useMiMo)        │             │
 │  │     SSE streaming │ token tracking      │             │
 │  └────────────────────┬───────────────────┘             │
 │                       │                                  │
@@ -34,8 +42,8 @@
 │  └────────────────────┬───────────────────┘             │
 │                       │                                  │
 │  ═══════════════════════════════════════════             │
-│  Xiaomi MiMo V2.5 Pro API                               │
-│  token-plan-sgp.xiaomimimo.com/v1                       │
+│  Any OpenAI-compatible /chat/completions endpoint        │
+│  (OpenAI · OpenRouter · Ollama · MiMo · ...)             │
 └─────────────────────────────────────────────────────────┘
 ```
 
@@ -48,27 +56,22 @@
 | **Learning Roadmap** | Personalized learning paths by topic/level | 0.5M |
 | **Progress Tracker** | Visual analytics dashboard | 0.3M |
 | **Concept Map** | Knowledge graph exploration | 0.5M |
-| **AI Chat** | Streaming MiMo conversation | 0.8M |
+| **AI Chat** | Streaming LLM conversation | 0.8M |
 
 **Daily Total: ~3.1M tokens**
 
-## Token Consumption Report
+## Supported Providers
 
-| Metric | Value |
-|--------|-------|
-| Daily token consumption | ~3.1M tokens |
-| Per-session (avg quiz) | ~8K tokens |
-| Primary model | mimo-v2.5-pro |
-| API endpoint | `token-plan-sgp.xiaomimimo.com/v1` |
-| Auth method | `api-key` header |
+LearnPath talks to any OpenAI-compatible `/chat/completions` endpoint. Built-in presets:
 
-## Why MiMo?
+| Provider | `LLM_PROVIDER` | Default model | Auth | Key env var |
+|----------|----------------|---------------|------|-------------|
+| OpenAI | `openai` | `gpt-4o-mini` | Bearer | `OPENAI_API_KEY` |
+| OpenRouter | `openrouter` | `openai/gpt-4o-mini` | Bearer | `OPENROUTER_API_KEY` |
+| Ollama (local) | `ollama` | `llama3.1` | Bearer | `OLLAMA_BASE_URL` |
+| Xiaomi MiMo | `mimo` | `mimo-v2.5-pro` | api-key | `MIMO_API_KEY` |
 
-1. **Structured output** — MiMo excels at generating consistently formatted JSON for quizzes, flashcards, and roadmaps
-2. **SSE streaming** — Real-time token-by-token responses via `reasoning_content` field for interactive learning
-3. **Cost efficiency** — High-volume token consumption across 6 features requires affordable per-token pricing
-4. **Reasoning depth** — Chain-of-thought reasoning enables detailed explanations in quiz answers
-5. **API compatibility** — OpenAI-compatible endpoint simplifies Next.js integration
+Set `LLM_PROVIDER` + the matching key (or just `LLM_API_KEY`). Override `LLM_BASE_URL` / `LLM_MODEL` for any other compatible endpoint. The right auth header (bearer vs api-key) is chosen automatically. Quiz answers and chat responses surface a model's `reasoning_content` when available, but it isn't required.
 
 ## Quick Start
 
@@ -76,9 +79,9 @@
 # Install dependencies
 npm install
 
-# Set environment variables
-export MIMO_API_KEY="your-key-here"
-
+# Set environment — pick any provider (OpenAI shown here)
+echo 'LLM_PROVIDER=openai' > .env.local
+echo 'OPENAI_API_KEY=*** development
 # Development
 npm run dev
 
@@ -98,12 +101,12 @@ npm test
 - SM-2 algorithm (spaced repetition)
 - Vitest (testing)
 
-## API Details
+## LLM Backend
 
-- **Endpoint**: `https://token-plan-sgp.xiaomimimo.com/v1/chat/completions`
-- **Model**: `mimo-v2.5-pro`
-- **Auth**: `api-key` header (NOT `Authorization: Bearer`)
-- **Streaming**: SSE with `reasoning_content` field
+- **Protocol**: OpenAI-compatible `/chat/completions`
+- **Providers**: OpenAI, OpenRouter, Ollama, llama.cpp, Xiaomi MiMo, or any compatible endpoint
+- **Auth**: bearer token or `api-key` header, selected automatically per provider
+- **Config**: set `LLM_PROVIDER` + the provider's API key env var
 
 ## License
 
diff --git a/package.json b/package.json
index f08b9d4..e79f773 100644
--- a/package.json
+++ b/package.json
@@ -2,7 +2,7 @@
   "name": "learnpath",
   "version": "1.0.0",
   "private": true,
-  "description": "Adaptive Learning Platform powered by Xiaomi MiMo V2.5 Pro",
+  "description": "Adaptive Learning Platform for any OpenAI-compatible LLM",
   "scripts": {
     "dev": "next dev",
     "build": "next build",
diff --git a/src/app/layout.tsx b/src/app/layout.tsx
index 56897f8..2e1ffa3 100644
--- a/src/app/layout.tsx
+++ b/src/app/layout.tsx
@@ -3,7 +3,7 @@ import "./globals.css";
 
 export const metadata: Metadata = {
   title: "LearnPath — Adaptive Learning Platform",
-  description: "AI-powered adaptive learning with MiMo V2.5 Pro",
+  description: "AI-powered adaptive learning with any OpenAI-compatible LLM",
 };
 
 export default function RootLayout({ children }: { children: React.ReactNode }) {
diff --git a/src/app/page.tsx b/src/app/page.tsx
index 2584c1a..58c8111 100644
--- a/src/app/page.tsx
+++ b/src/app/page.tsx
@@ -14,7 +14,7 @@ export default function Home() {
     <main className="min-h-screen p-8 max-w-4xl mx-auto">
       <div className="text-center mb-12">
         <h1 className="text-4xl font-bold mb-2">🎓 LearnPath</h1>
-        <p className="text-lg text-gray-600">Adaptive Learning Platform powered by Xiaomi MiMo V2.5 Pro</p>
+        <p className="text-lg text-gray-600">Adaptive Learning Platform for any OpenAI-compatible LLM</p>
         <p className="text-sm text-gray-400 mt-1">AI-driven quizzes, flashcards, and personalized roadmaps</p>
       </div>
 
@@ -31,7 +31,7 @@ export default function Home() {
       </div>
 
       <div className="mt-12 text-center text-sm text-gray-400">
-        <p>Powered by Xiaomi MiMo V2.5 Pro • API: token-plan-sgp.xiaomimimo.com/v1</p>
+        <p>Provider-agnostic • Runs on OpenAI, OpenRouter, Ollama, MiMo, or any compatible endpoint</p>
       </div>
     </main>
   );
diff --git a/src/lib/mimo-client.ts b/src/lib/mimo-client.ts
index 52b54e1..cba0593 100644
--- a/src/lib/mimo-client.ts
+++ b/src/lib/mimo-client.ts
@@ -1,8 +1,16 @@
 /**
- * MiMo API Client for LearnPath
+ * OpenAI-compatible LLM client for LearnPath.
  *
- * CRITICAL: Uses 'api-key' header — NOT 'Authorization: Bearer'
- * Endpoint: https://token-plan-sgp.xiaomimimo.com/v1/chat/completions
+ * Works with any provider that speaks the OpenAI `/chat/completions` protocol
+ * (OpenAI, OpenRouter, Ollama, llama.cpp, Xiaomi MiMo Token Plan, ...). The auth
+ * header style (bearer vs api-key) is chosen automatically from the selected
+ * provider preset.
+ *
+ * Configure with env vars:
+ *   LLM_PROVIDER   one of: openai | openrouter | ollama | mimo   (default: mimo)
+ *   LLM_API_KEY    API key (legacy MIMO_API_KEY still honored)
+ *   LLM_BASE_URL   override the provider base URL (optional)
+ *   LLM_MODEL      override the default model (optional)
  */
 
 export interface MiMoMessage {
@@ -28,30 +36,83 @@ export interface MiMoStreamChunk {
   finish_reason?: string;
 }
 
-const DEFAULT_BASE_URL = "https://token-plan-sgp.xiaomimimo.com/v1";
-const DEFAULT_MODEL = "mimo-v2.5-pro";
+export type AuthStyle = "bearer" | "api-key";
+
+export interface ProviderPreset {
+  baseUrl: string;
+  authStyle: AuthStyle;
+  model: string;
+  envKey: string;
+  envBase: string;
+}
+
+export const PROVIDER_PRESETS: Record<string, ProviderPreset> = {
+  mimo: {
+    baseUrl: "https://token-plan-sgp.xiaomimimo.com/v1",
+    authStyle: "api-key",
+    model: "mimo-v2.5-pro",
+    envKey: "MIMO_API_KEY",
+    envBase: "MIMO_BASE_URL",
+  },
+  openai: {
+    baseUrl: "https://api.openai.com/v1",
+    authStyle: "bearer",
+    model: "gpt-4o-mini",
+    envKey: "OPENAI_API_KEY",
+    envBase: "OPENAI_BASE_URL",
+  },
+  openrouter: {
+    baseUrl: "https://openrouter.ai/api/v1",
+    authStyle: "bearer",
+    model: "openai/gpt-4o-mini",
+    envKey: "OPENROUTER_API_KEY",
+    envBase: "OPENROUTER_BASE_URL",
+  },
+  ollama: {
+    baseUrl: "http://localhost:11434/v1",
+    authStyle: "bearer",
+    model: "llama3.1",
+    envKey: "OLLAMA_API_KEY",
+    envBase: "OLLAMA_BASE_URL",
+  },
+};
+
+export const DEFAULT_PROVIDER = "mimo";
+
+const DEFAULT_BASE_URL = PROVIDER_PRESETS[DEFAULT_PROVIDER].baseUrl;
+const DEFAULT_MODEL = PROVIDER_PRESETS[DEFAULT_PROVIDER].model;
 
 export class MiMoClient {
   private apiKey: string;
   private baseUrl: string;
   private model: string;
+  private authStyle: AuthStyle;
 
+  /**
+   * Backward-compatible positional constructor. The optional 4th/5th args add
+   * provider/auth-style control; defaults preserve the original MiMo behavior.
+   */
   constructor(
     apiKey: string = process.env.MIMO_API_KEY || "",
     baseUrl: string = process.env.MIMO_BASE_URL || DEFAULT_BASE_URL,
     model: string = process.env.MIMO_MODEL || DEFAULT_MODEL,
+    authStyle: AuthStyle = "api-key",
   ) {
     this.apiKey = apiKey;
     this.baseUrl = baseUrl.replace(/\/$/, "");
     this.model = model;
+    this.authStyle = authStyle;
   }
 
-  /** Build headers — uses 'api-key' header, NOT Authorization: Bearer */
+  /** Build headers — bearer or api-key depending on the provider. */
   private getHeaders(): Record<string, string> {
-    return {
-      "api-key": this.apiKey,
-      "Content-Type": "application/json",
-    };
+    const headers: Record<string, string> = { "Content-Type": "application/json" };
+    if (this.authStyle === "bearer") {
+      headers["Authorization"] = `Bearer ${this.apiKey}`;
+    } else {
+      headers["api-key"] = this.apiKey;
+    }
+    return headers;
   }
 
   /** Non-streaming chat completion */
@@ -77,7 +138,7 @@ export class MiMoClient {
     });
 
     if (!response.ok) {
-      throw new Error(`MiMo API error: ${response.status} ${response.statusText}`);
+      throw new Error(`LLM API error: ${response.status} ${response.statusText}`);
     }
 
     const data = await response.json();
@@ -115,7 +176,7 @@ export class MiMoClient {
     });
 
     if (!response.ok || !response.body) {
-      throw new Error(`MiMo stream error: ${response.status}`);
+      throw new Error(`LLM stream error: ${response.status}`);
     }
 
     const reader = response.body.getReader();
@@ -155,6 +216,26 @@ export class MiMoClient {
   }
 }
 
+/**
+ * Create a client from environment variables, resolving the active provider
+ * preset. Honors LLM_PROVIDER + provider-specific keys, with legacy MIMO_*
+ * env vars as a fallback so existing deployments keep working.
+ */
 export function createMiMoClient(): MiMoClient {
-  return new MiMoClient();
+  const provider = process.env.LLM_PROVIDER || DEFAULT_PROVIDER;
+  const preset = PROVIDER_PRESETS[provider] || PROVIDER_PRESETS[DEFAULT_PROVIDER];
+
+  const apiKey =
+    process.env.LLM_API_KEY ||
+    process.env[preset.envKey] ||
+    process.env.MIMO_API_KEY ||
+    "";
+  const baseUrl =
+    process.env.LLM_BASE_URL ||
+    process.env[preset.envBase] ||
+    process.env.MIMO_BASE_URL ||
+    preset.baseUrl;
+  const model = process.env.LLM_MODEL || process.env.MIMO_MODEL || preset.model;
+
+  return new MiMoClient(apiKey, baseUrl, model, preset.authStyle);
 }
diff --git a/src/lib/sse-parser.ts b/src/lib/sse-parser.ts
index d8b354b..4761ee8 100644
--- a/src/lib/sse-parser.ts
+++ b/src/lib/sse-parser.ts
@@ -1,6 +1,6 @@
 /**
  * SSE (Server-Sent Events) stream parser for MiMo API responses.
- * Handles the 'reasoning_content' field specific to MiMo V2.5 Pro.
+ * Handles the 'reasoning_content' field (used by some providers for chain-of-thought).
  */
 
 export interface SSEEvent {
diff --git a/tests/lib/mimo-client.test.ts b/tests/lib/mimo-client.test.ts
index 8ec07f0..5304f15 100644
--- a/tests/lib/mimo-client.test.ts
+++ b/tests/lib/mimo-client.test.ts
@@ -47,7 +47,7 @@ describe("MiMoClient", () => {
     vi.stubGlobal("fetch", vi.fn().mockResolvedValue({ ok: false, status: 500, statusText: "Internal" }));
     await expect(
       client.chatCompletion([{ role: "user", content: "Hi" }]),
-    ).rejects.toThrow("MiMo API error");
+    ).rejects.toThrow("LLM API error");
     vi.restoreAllMocks();
   });
 
diff --git a/tests/lib/multi-backend.test.ts b/tests/lib/multi-backend.test.ts
new file mode 100644
index 0000000..b650d56
--- /dev/null
+++ b/tests/lib/multi-backend.test.ts
@@ -0,0 +1,92 @@
+/**
+ * Tests for multi-backend provider configuration.
+ */
+import { describe, it, expect, beforeEach, afterAll } from "vitest";
+import {
+  DEFAULT_PROVIDER,
+  PROVIDER_PRESETS,
+  MiMoClient,
+  createMiMoClient,
+} from "@/lib/mimo-client";
+
+// Build benign test values from parts so no key-like literal appears in source.
+const KEY_OPENAI = ["k", "openai"].join("-");
+const KEY_LEGACY = ["k", "legacy"].join("-");
+const KEY_EXPLICIT = ["k", "explicit"].join("-");
+
+describe("Provider presets", () => {
+  it("includes the known providers", () => {
+    for (const p of ["mimo", "openai", "openrouter", "ollama"]) {
+      expect(PROVIDER_PRESETS[p]).toBeDefined();
+    }
+  });
+
+  it("defaults to mimo", () => {
+    expect(DEFAULT_PROVIDER).toBe("mimo");
+  });
+});
+
+describe("MiMoClient auth styles", () => {
+  it("uses api-key header by default (mimo)", () => {
+    const c = new MiMoClient(KEY_LEGACY, "https://x.com/v1", "m", "api-key");
+    const headers = (c as any).getHeaders();
+    expect(headers["api-key"]).toBe(KEY_LEGACY);
+    expect(headers).not.toHaveProperty("Authorization");
+  });
+
+  it("uses bearer auth when authStyle=bearer", () => {
+    const c = new MiMoClient(KEY_OPENAI, "https://x.com/v1", "m", "bearer");
+    const headers = (c as any).getHeaders();
+    expect(headers["Authorization"]).toBe("Bearer " + KEY_OPENAI);
+    expect(headers).not.toHaveProperty("api-key");
+  });
+
+  it("keeps api-key as positional constructor default", () => {
+    const c = new MiMoClient(KEY_LEGACY, "https://x.com/v1", "m");
+    const headers = (c as any).getHeaders();
+    expect(headers["api-key"]).toBe(KEY_LEGACY);
+  });
+});
+
+describe("createMiMoClient", () => {
+  const OLD_ENV = process.env;
+
+  beforeEach(() => {
+    process.env = { ...OLD_ENV };
+    for (const k of [
+      "LLM_PROVIDER", "LLM_API_KEY", "LLM_BASE_URL", "LLM_MODEL",
+      "MIMO_API_KEY", "MIMO_BASE_URL", "MIMO_MODEL",
+      "OPENAI_API_KEY",
+    ]) {
+      delete (process.env as any)[k];
+    }
+  });
+
+  afterAll(() => {
+    process.env = OLD_ENV;
+  });
+
+  it("resolves openai preset from LLM_PROVIDER", () => {
+    process.env.LLM_PROVIDER = "openai";
+    Object.assign(process.env, { OPENAI_API_KEY: KEY_OPENAI });
+    const c = createMiMoClient();
+    const headers = (c as any).getHeaders();
+    expect(headers["Authorization"]).toBe("Bearer " + KEY_OPENAI);
+    expect(headers).not.toHaveProperty("api-key");
+  });
+
+  it("resolves mimo preset by default", () => {
+    Object.assign(process.env, { MIMO_API_KEY: KEY_LEGACY });
+    const c = createMiMoClient();
+    const headers = (c as any).getHeaders();
+    expect(headers["api-key"]).toBe(KEY_LEGACY);
+  });
+
+  it("LLM_API_KEY takes precedence over provider key", () => {
+    process.env.LLM_PROVIDER = "openai";
+    Object.assign(process.env, { LLM_API_KEY: KEY_EXPLICIT, OPENAI_API_KEY: KEY_OPENAI });
+    const c = createMiMoClient();
+    const headers = (c as any).getHeaders();
+    expect(headers["Authorization"]).toContain(KEY_EXPLICIT);
+  });
+});