로컬 Ollama와 이미지 생성을 지원하는 비공개 오프라인 멀티모델 채팅 클라이언트

I can't execute commands or inspect your repo from here, but here's a complete, copy-pasteable setup guide you can run locally. I'll flag the one thing I can't verify (your repo's existing config) so you can check it first.

## 0) Check your repo for existing config

Before anything else, look for existing LLM X config so you don't duplicate or conflict:

```bash
# from your repo root
ls -la
find . -maxdepth 3 -iname "*llm*" -o -iname "*.env*" -o -iname "config*" | grep -v node_modules
grep -rIl --exclude-dir=node_modules -e "OLLAMA" -e "LM Studio" -e "AUTOMATIC1111" -e "localhost:11434" -e "localhost:1234" .
```

Note what you find (env var names, ports, provider keys). If there's an existing `.env.example` or config file, align the steps below with it rather than inventing new names.

## 1) Confirm Ollama is installed and pull a small model

```bash
ollama --version
# if missing: install from https://ollama.com/download (or `brew install ollama` on macOS)

# pull a small model (pick one)
ollama pull llama3.2:1b        # ~1.3 GB, fast
# or
ollama pull qwen2.5:0.5b       # ~400 MB, very small
```

## 2) Start Ollama with OLLAMA_ORIGINS set

Ollama must allow your web app origin and the Chrome extension origin. Chrome extension origins look like `chrome-extension://<extension-id>`.

```bash
# stop any running instance first
pkill ollama 2>/dev/null || true

export OLLAMA_ORIGINS="https://mrdjohnson.github.io,chrome-extension://YOUR_EXTENSION_ID"
# add localhost origins too if you test the app locally:
# export OLLAMA_ORIGINS="https://mrdjohnson.github.io,chrome-extension://YOUR_EXTENSION_ID,http://localhost:3000,http://127.0.0.1:3000"

ollama serve
```

Replace `YOUR_EXTENSION_ID` with the actual ID from `chrome://extensions` (Developer mode on). If you don't know it yet, you can temporarily use `*` for testing, but lock it down afterward:

```bash
# temporary, permissive — for debugging only
export OLLAMA_ORIGINS="*"
```

Verify it's up:

```bash
curl http://localhost:11434/api/tags
```

## 3) Optional: LM Studio (CORS) and AUTOMATIC1111 (API)

**LM Studio** — enable CORS in the UI: *Developer* tab → toggle **Enable CORS**, then **Start Server** (default `http://localhost:1234`). Verify:

```bash
curl http://localhost:1234/v1/models
```

**AUTOMATIC1111** — start with API + CORS:

```bash
cd /path/to/stable-diffusion-webui
./webui.sh --api --listen --cors-allow-origins "*"
# Windows: webui-user.bat with COMMANDLINE_ARGS="--api --listen --cors-allow-origins *"
```

Verify:

```bash
curl http://localhost:7860/sdapi/v1/sd-models
```

## 4) Verify the app connects and chats with two models at once

In the LLM X UI:

1. Add an **Ollama** provider pointing at `http://localhost:11434`.
2. Add a second provider (LM Studio at `http://localhost:1234/v1`, or a second Ollama model).
3. Open a chat, select **two models** (multi-model / side-by-side mode), send one prompt, and confirm both respond.

Quick sanity check that both backends answer independently:

```bash
curl http://localhost:11434/api/generate -d '{"model":"llama3.2:1b","prompt":"hi","stream":false}'
curl http://localhost:1234/v1/chat/completions -H "Content-Type: application/json" \
  -d '{"model":"your-lmstudio-model","messages":[{"role":"user","content":"hi"}]}'
```

If the browser app fails but curl works, it's almost always `OLLAMA_ORIGINS` (or LM Studio CORS) — recheck the exact origin string, including scheme and port.

## 5) README section to add

```markdown
## Local LLM Setup (Ollama + LM Studio + AUTOMATIC1111)

### Prerequisites
- Ollama installed (`ollama --version`)
- Optional: LM Studio, AUTOMATIC1111

### 1. Pull a small model
```bash
ollama pull llama3.2:1b
```

### 2. Run Ollama with allowed origins
```bash
pkill ollama 2>/dev/null || true
export OLLAMA_ORIGINS="https://mrdjohnson.github.io,chrome-extension://YOUR_EXTENSION_ID"
ollama serve
```
Replace `YOUR_EXTENSION_ID` with the ID from `chrome://extensions`.

### 3. Optional services
- **LM Studio:** Developer tab → Enable CORS → Start Server (`http://localhost:1234`).
- **AUTOMATIC1111:**
  ```bash
  ./webui.sh --api --listen --cors-allow-origins "*"
  ```

### 4. Verify
```bash
curl http://localhost:11434/api/tags
curl http://localhost:1234/v1/models
curl http://localhost:7860/sdapi/v1/sd-models
```
Then in the app, add the Ollama provider (`http://localhost:11434`), add a second provider/model, and confirm a two-model chat responds from both.

### Troubleshooting
- Browser blocked but curl works → check `OLLAMA_ORIGINS` / LM Studio CORS origin strings exactly.
- Extension can't connect → confirm `chrome-extension://<id>` is in `OLLAMA_ORIGINS`.
```

Two things to confirm on your side: the actual **extension ID** (I used a placeholder), and whether your repo already defines provider config/env names in step 0 — if so, match those instead of the defaults above.