diff --git a/.claude/settings.json b/.claude/settings.json
new file mode 100644
index 0000000..0f243d7
--- /dev/null
+++ b/.claude/settings.json
@@ -0,0 +1,5 @@
+{
+  "enabledPlugins": {
+    "superpowers@claude-plugins-official": true
+  }
+}
diff --git a/CLAUDE.md b/CLAUDE.md
new file mode 100644
index 0000000..68a08fe
--- /dev/null
+++ b/CLAUDE.md
@@ -0,0 +1,83 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+
+## Project Overview
+
+CoinTrader is a Python asyncio-based automated cryptocurrency trading bot for Binance Futures. It trades XRPUSDT on 15-minute candles, using BTC/ETH as correlation features. The system has 5 layers: Data (WebSocket streams) → Signal (technical indicators) → ML Filter (ONNX/LightGBM) → Execution & Risk → Event/Alert (Discord).
+
+## Common Commands
+
+```bash
+# venv 
+source .venv/bin/activate
+
+# Run the bot
+python main.py
+
+# Run full test suite
+bash scripts/run_tests.sh
+
+# Run filtered tests
+bash scripts/run_tests.sh -k "bot"
+
+# Run pytest directly
+pytest tests/ -v --tb=short
+
+# ML training pipeline (LightGBM default)
+bash scripts/train_and_deploy.sh
+
+# MLX GPU training (macOS Apple Silicon)
+bash scripts/train_and_deploy.sh mlx
+
+# Hyperparameter tuning (50 trials, 5-fold walk-forward)
+python scripts/tune_hyperparams.py
+
+# Fetch historical data
+python scripts/fetch_history.py --symbols XRPUSDT BTCUSDT ETHUSDT --interval 15m --days 365
+
+# Deploy models to production
+bash scripts/deploy_model.sh
+```
+
+## Architecture
+
+**Entry point**: `main.py` → creates `Config` (dataclass from env vars) → runs `TradingBot`
+
+**5-layer data flow on each 15m candle close:**
+1. `src/data_stream.py` — Combined WebSocket for XRP/BTC/ETH, deque buffers (200 candles each)
+2. `src/indicators.py` — RSI, MACD, BB, EMA, StochRSI, ATR; weighted signal aggregation → LONG/SHORT/HOLD
+3. `src/ml_filter.py` + `src/ml_features.py` — 23-feature extraction, ONNX priority > LightGBM fallback, threshold ≥ 0.60
+4. `src/exchange.py` + `src/risk_manager.py` — Dynamic margin, MARKET orders with SL/TP, daily loss limit (5%)
+5. `src/user_data_stream.py` + `src/notifier.py` — Real-time TP/SL detection via WebSocket, Discord webhooks
+
+**Parallel execution**: `user_data_stream` runs independently via `asyncio.gather()` alongside candle processing.
+
+## Key Patterns
+
+- **Async-first**: All I/O operations use `async/await`; parallel tasks via `asyncio.gather()`
+- **Reverse signal re-entry**: While holding LONG, if SHORT signal appears → close position, cancel SL/TP, open SHORT. `_is_reentering` flag prevents race conditions with User Data Stream
+- **ML hot reload**: `ml_filter.check_and_reload()` compares file mtime on every candle, reloads model without restart
+- **Active Config pattern**: Best hyperparams stored in `models/active_lgbm_params.json`, must be manually approved before retraining
+- **Graceful degradation**: Missing model → all signals pass; API failure → use fallback values (0.0 for OI/funding)
+- **Walk-forward validation**: Time-series CV with undersampling (1:1 class balance, preserving time order)
+- **Label generation**: Binary labels based on 24-candle (6h) lookahead — check SL hit first (conservative), then TP
+
+## Testing
+
+- All external APIs (Binance, Discord) are mocked with `unittest.mock.AsyncMock`
+- Async tests use `@pytest.mark.asyncio`
+- 14 test files, 80+ test cases covering all layers
+- Testing is done in actual terminal, not IDE sandbox
+
+## Configuration
+
+Environment variables via `.env` file (see `.env.example`). Key vars: `BINANCE_API_KEY`, `BINANCE_API_SECRET`, `SYMBOL` (default XRPUSDT), `LEVERAGE`, `DISCORD_WEBHOOK_URL`, `MARGIN_MAX_RATIO`, `MARGIN_MIN_RATIO`, `NO_ML_FILTER`.
+
+`src/config.py` uses `@dataclass` with `__post_init__` to load and validate all env vars.
+
+## Deployment
+
+- **Docker**: `Dockerfile` (Python 3.12-slim) + `docker-compose.yml`
+- **CI/CD**: Jenkins pipeline (Gitea → Docker registry → LXC production server)
+- Models stored in `models/`, data cache in `data/`, logs in `logs/`
diff --git a/docs/plans/2026-03-02-adx-filter-design.md b/docs/plans/2026-03-02-adx-filter-design.md
new file mode 100644
index 0000000..863f549
--- /dev/null
+++ b/docs/plans/2026-03-02-adx-filter-design.md
@@ -0,0 +1,150 @@
+# ADX 횡보장 필터 구현 계획
+
+> **For Claude:** REQUIRED SUB-SKILL: Use superpowers:executing-plans to implement this plan task-by-task.
+
+**Goal:** ADX < 25일 때 get_signal()에서 즉시 HOLD를 반환하여 횡보장 진입을 차단한다.
+
+**Architecture:** `calculate_all()`에서 `pandas_ta.adx()`로 ADX 컬럼을 추가하고, `get_signal()`에서 가중치 계산 전 ADX < 25이면 early-return HOLD. NaN(초기 캔들)은 기존 로직으로 폴백.
+
+**Tech Stack:** pandas-ta (이미 사용 중), pytest
+
+---
+
+### Task 1: ADX 계산 테스트 추가
+
+**Files:**
+- Test: `tests/test_indicators.py`
+
+**Step 1: Write the failing test**
+
+```python
+def test_adx_column_exists(sample_df):
+    """calculate_all()이 adx 컬럼을 생성하는지 확인."""
+    ind = Indicators(sample_df)
+    df = ind.calculate_all()
+    assert "adx" in df.columns
+    valid = df["adx"].dropna()
+    assert (valid >= 0).all()
+```
+
+`tests/test_indicators.py`에 위 테스트 함수를 추가한다.
+
+**Step 2: Run test to verify it fails**
+
+Run: `pytest tests/test_indicators.py::test_adx_column_exists -v`
+Expected: FAIL — `"adx" not in df.columns`
+
+---
+
+### Task 2: calculate_all()에 ADX 계산 추가
+
+**Files:**
+- Modify: `src/indicators.py:46-48` (vol_ma20 계산 바로 앞에 추가)
+
+**Step 3: Write minimal implementation**
+
+`calculate_all()`의 Stochastic RSI 계산 뒤, `vol_ma20` 계산 앞에 추가:
+
+```python
+        # ADX (14) — 횡보장 필터
+        adx_df = ta.adx(df["high"], df["low"], df["close"], length=14)
+        df["adx"] = adx_df["ADX_14"]
+```
+
+**Step 4: Run test to verify it passes**
+
+Run: `pytest tests/test_indicators.py::test_adx_column_exists -v`
+Expected: PASS
+
+**Step 5: Commit**
+
+```bash
+git add src/indicators.py tests/test_indicators.py
+git commit -m "feat: add ADX calculation to indicators"
+```
+
+---
+
+### Task 3: ADX 필터 테스트 추가 (차단 케이스)
+
+**Files:**
+- Test: `tests/test_indicators.py`
+
+**Step 6: Write the failing test**
+
+```python
+def test_adx_filter_blocks_low_adx(sample_df):
+    """ADX < 25일 때 가중치와 무관하게 HOLD를 반환해야 한다."""
+    ind = Indicators(sample_df)
+    df = ind.calculate_all()
+    # ADX를 강제로 낮은 값으로 설정
+    df["adx"] = 15.0
+    signal = ind.get_signal(df)
+    assert signal == "HOLD"
+```
+
+**Step 7: Run test to verify it fails**
+
+Run: `pytest tests/test_indicators.py::test_adx_filter_blocks_low_adx -v`
+Expected: FAIL — signal이 LONG 또는 SHORT 반환 (ADX 필터 미구현)
+
+---
+
+### Task 4: ADX 필터 테스트 추가 (NaN 폴백 케이스)
+
+**Files:**
+- Test: `tests/test_indicators.py`
+
+**Step 8: Write the failing test**
+
+```python
+def test_adx_nan_falls_through(sample_df):
+    """ADX가 NaN(초기 캔들)이면 기존 가중치 로직으로 폴백해야 한다."""
+    ind = Indicators(sample_df)
+    df = ind.calculate_all()
+    df["adx"] = float("nan")
+    signal = ind.get_signal(df)
+    # NaN이면 차단하지 않고 기존 로직 실행 → LONG/SHORT/HOLD 중 하나
+    assert signal in ("LONG", "SHORT", "HOLD")
+```
+
+**Step 9: Run test to verify it passes (이 테스트는 현재도 통과)**
+
+Run: `pytest tests/test_indicators.py::test_adx_nan_falls_through -v`
+Expected: PASS (ADX 컬럼이 무시되므로 기존 로직 그대로)
+
+---
+
+### Task 5: get_signal()에 ADX early-return 구현
+
+**Files:**
+- Modify: `src/indicators.py:51-56` (get_signal 메서드 시작부)
+
+**Step 10: Write minimal implementation**
+
+`get_signal()` 메서드의 `last = df.iloc[-1]` 바로 다음에 추가:
+
+```python
+        # ADX 횡보장 필터: ADX < 25이면 추세 부재로 판단하여 진입 차단
+        adx = last.get("adx", None)
+        if adx is not None and not pd.isna(adx) and adx < 25:
+            logger.debug(f"ADX 필터: {adx:.1f} < 25 — HOLD")
+            return "HOLD"
+```
+
+**Step 11: Run all ADX-related tests**
+
+Run: `pytest tests/test_indicators.py -k "adx" -v`
+Expected: 3 tests PASS
+
+**Step 12: Run full test suite to check for regressions**
+
+Run: `pytest tests/ -v --tb=short`
+Expected: All tests PASS
+
+**Step 13: Commit**
+
+```bash
+git add src/indicators.py tests/test_indicators.py
+git commit -m "feat: add ADX filter to block sideways market entries"
+```