feat(benchmark): add hook mode with baseline cache for pre-push checks

- Fix lint errors and LSP issues in benchmark.py - Add --hook mode to compare against saved baseline - Add --baseline flag to save results as baseline - Add --threshold to configure degradation threshold (default 20%) - Add benchmark step to pre-push hook in hk.pkl - Update AGENTS.md with hk documentation links and benchmark runner docs
refactor: modularize display backends and add benchmark runner
2026-03-15 22:41:13 -07:00 · 2026-03-15 22:25:28 -07:00 · 2026-03-15 22:13:44 -07:00
14 changed files with 1249 additions and 163 deletions
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -36,12 +36,13 @@ mise run ci             # Full CI pipeline (topics-init + lint + test-cov)

 ```bash
 mise run run            # Run mainline (terminal)
-mise run run-poetry     # Run with poetry feed
-mise run run-firehose   # Run in firehose mode
-mise run run-websocket  # Run with WebSocket display only
-mise run run-both       # Run with both terminal and WebSocket
-mise run run-client     # Run both + open browser
-mise run cmd            # Run C&C command interface
+mise run run-poetry    # Run with poetry feed
+mise run run-firehose  # Run in firehose mode
+mise run run-websocket # Run with WebSocket display only
+mise run run-sixel     # Run with Sixel graphics display
+mise run run-both      # Run with both terminal and WebSocket
+mise run run-client    # Run both + open browser
+mise run cmd           # Run C&C command interface
 ```

 ## Git Hooks
@@ -59,9 +60,52 @@ hk init --mise
 mise run pre-commit
 ```

+**IMPORTANT**: Always review the hk documentation before modifying `hk.pkl`:
+- [hk Configuration Guide](https://hk.jdx.dev/configuration.html)
+- [hk Hooks Reference](https://hk.jdx.dev/hooks.html)
+- [hk Builtins](https://hk.jdx.dev/builtins.html)
+
 The project uses hk configured in `hk.pkl`:
 - **pre-commit**: runs ruff-format and ruff (with auto-fix)
- **pre-push**: runs ruff check
+- **pre-push**: runs ruff check + benchmark hook
+
+## Benchmark Runner
+
+Run performance benchmarks:
+
+```bash
+mise run benchmark           # Run all benchmarks (text output)
+mise run benchmark-json     # Run benchmarks (JSON output)
+mise run benchmark-report   # Run benchmarks (Markdown report)
+```
+
+### Benchmark Commands
+
+```bash
+# Run benchmarks
+uv run python -m engine.benchmark
+
+# Run with specific displays/effects
+uv run python -m engine.benchmark --displays null,terminal --effects fade,glitch
+
+# Save baseline for hook comparisons
+uv run python -m engine.benchmark --baseline
+
+# Run in hook mode (compares against baseline)
+uv run python -m engine.benchmark --hook
+
+# Hook mode with custom threshold (default: 20% degradation)
+uv run python -m engine.benchmark --hook --threshold 0.3
+
+# Custom baseline location
+uv run python -m engine.benchmark --hook --cache /path/to/cache.json
+```
+
+### Hook Mode
+
+The `--hook` mode compares current benchmarks against a saved baseline. If performance degrades beyond the threshold (default 20%), it exits with code 1. This is useful for preventing performance regressions in feature branches.
+
+The pre-push hook runs benchmark in hook mode to catch performance regressions before pushing.

 ## Workflow Rules

@@ -128,6 +172,7 @@ The project uses pytest with strict marker enforcement. Test configuration is in
 - **Display abstraction** (`engine/display.py`): swap display backends via the Display protocol
  - `TerminalDisplay` - ANSI terminal output
  - `WebSocketDisplay` - broadcasts to web clients via WebSocket
+  - `SixelDisplay` - renders to Sixel graphics (pure Python, no C dependency)
  - `MultiDisplay` - forwards to multiple displays simultaneously

 - **WebSocket display** (`engine/websocket_display.py`): real-time frame broadcasting to web browsers
@@ -135,6 +180,12 @@ The project uses pytest with strict marker enforcement. Test configuration is in
  - HTTP server on port 8766 (serves HTML client)
  - Client at `client/index.html` with ANSI color parsing and fullscreen support

+- **Display modes** (`--display` flag):
+  - `terminal` - Default ANSI terminal output
+  - `websocket` - Web browser display (requires websockets package)
+  - `sixel` - Sixel graphics in supported terminals (iTerm2, mintty, etc.)
+  - `both` - Terminal + WebSocket simultaneously
+
 ### Command & Control

 - C&C uses separate ntfy topics for commands and responses
--- a/engine/benchmark.py
+++ b/engine/benchmark.py
@@ -0,0 +1,638 @@
+#!/usr/bin/env python3
+"""
+Benchmark runner for mainline - tests performance across effects and displays.
+
+Usage:
+    python -m engine.benchmark
+    python -m engine.benchmark --output report.md
+    python -m engine.benchmark --displays terminal,websocket --effects glitch,fade
+    python -m engine.benchmark --format json --output benchmark.json
+
+Headless mode (default): suppress all terminal output during benchmarks.
+"""
+
+import argparse
+import json
+import sys
+import time
+from dataclasses import dataclass, field
+from datetime import datetime
+from io import StringIO
+from pathlib import Path
+from typing import Any
+
+import numpy as np
+
+
+@dataclass
+class BenchmarkResult:
+    """Result of a single benchmark run."""
+
+    name: str
+    display: str
+    effect: str | None
+    iterations: int
+    total_time_ms: float
+    avg_time_ms: float
+    std_dev_ms: float
+    min_ms: float
+    max_ms: float
+    fps: float
+    chars_processed: int
+    chars_per_sec: float
+
+
+@dataclass
+class BenchmarkReport:
+    """Complete benchmark report."""
+
+    timestamp: str
+    python_version: str
+    results: list[BenchmarkResult] = field(default_factory=list)
+    summary: dict[str, Any] = field(default_factory=dict)
+
+
+def get_sample_buffer(width: int = 80, height: int = 24) -> list[str]:
+    """Generate a sample buffer for benchmarking."""
+    lines = []
+    for i in range(height):
+        line = f"\x1b[32mLine {i}\x1b[0m " + "A" * (width - 10)
+        lines.append(line)
+    return lines
+
+
+def benchmark_display(
+    display_class, buffer: list[str], iterations: int = 100
+) -> BenchmarkResult | None:
+    """Benchmark a single display."""
+    old_stdout = sys.stdout
+    old_stderr = sys.stderr
+
+    try:
+        sys.stdout = StringIO()
+        sys.stderr = StringIO()
+
+        display = display_class()
+        display.init(80, 24)
+
+        times = []
+        chars = sum(len(line) for line in buffer)
+
+        for _ in range(iterations):
+            t0 = time.perf_counter()
+            display.show(buffer)
+            elapsed = (time.perf_counter() - t0) * 1000
+            times.append(elapsed)
+
+        display.cleanup()
+
+    except Exception:
+        return None
+    finally:
+        sys.stdout = old_stdout
+        sys.stderr = old_stderr
+
+    times_arr = np.array(times)
+
+    return BenchmarkResult(
+        name=f"display_{display_class.__name__}",
+        display=display_class.__name__,
+        effect=None,
+        iterations=iterations,
+        total_time_ms=sum(times),
+        avg_time_ms=float(np.mean(times_arr)),
+        std_dev_ms=float(np.std(times_arr)),
+        min_ms=float(np.min(times_arr)),
+        max_ms=float(np.max(times_arr)),
+        fps=float(1000.0 / np.mean(times_arr)) if np.mean(times_arr) > 0 else 0.0,
+        chars_processed=chars * iterations,
+        chars_per_sec=float((chars * iterations) / (sum(times) / 1000))
+        if sum(times) > 0
+        else 0.0,
+    )
+
+
+def benchmark_effect_with_display(
+    effect_class, display, buffer: list[str], iterations: int = 100
+) -> BenchmarkResult | None:
+    """Benchmark an effect with a display."""
+    old_stdout = sys.stdout
+    old_stderr = sys.stderr
+
+    try:
+        sys.stdout = StringIO()
+        sys.stderr = StringIO()
+
+        effect = effect_class()
+        effect.configure(enabled=True, intensity=1.0)
+
+        times = []
+        chars = sum(len(line) for line in buffer)
+
+        for _ in range(iterations):
+            processed = effect.process(buffer)
+            t0 = time.perf_counter()
+            display.show(processed)
+            elapsed = (time.perf_counter() - t0) * 1000
+            times.append(elapsed)
+
+        display.cleanup()
+
+    except Exception:
+        return None
+    finally:
+        sys.stdout = old_stdout
+        sys.stderr = old_stderr
+
+    times_arr = np.array(times)
+
+    return BenchmarkResult(
+        name=f"effect_{effect_class.__name__}_with_{display.__class__.__name__}",
+        display=display.__class__.__name__,
+        effect=effect_class.__name__,
+        iterations=iterations,
+        total_time_ms=sum(times),
+        avg_time_ms=float(np.mean(times_arr)),
+        std_dev_ms=float(np.std(times_arr)),
+        min_ms=float(np.min(times_arr)),
+        max_ms=float(np.max(times_arr)),
+        fps=float(1000.0 / np.mean(times_arr)) if np.mean(times_arr) > 0 else 0.0,
+        chars_processed=chars * iterations,
+        chars_per_sec=float((chars * iterations) / (sum(times) / 1000))
+        if sum(times) > 0
+        else 0.0,
+    )
+
+
+def get_available_displays():
+    """Get available display classes."""
+    from engine.display import (
+        DisplayRegistry,
+        NullDisplay,
+        TerminalDisplay,
+    )
+
+    DisplayRegistry.initialize()
+
+    displays = [
+        ("null", NullDisplay),
+        ("terminal", TerminalDisplay),
+    ]
+
+    try:
+        from engine.display.backends.websocket import WebSocketDisplay
+
+        displays.append(("websocket", WebSocketDisplay))
+    except Exception:
+        pass
+
+    try:
+        from engine.display.backends.sixel import SixelDisplay
+
+        displays.append(("sixel", SixelDisplay))
+    except Exception:
+        pass
+
+    return displays
+
+
+def get_available_effects():
+    """Get available effect classes."""
+    try:
+        from engine.effects import get_registry
+    except Exception:
+        return []
+
+    effects = []
+    registry = get_registry()
+
+    for name, effect in registry.list_all().items():
+        if effect:
+            effects.append((name, effect))
+
+    return effects
+
+
+def run_benchmarks(
+    displays: list[tuple[str, Any]] | None = None,
+    effects: list[tuple[str, Any]] | None = None,
+    iterations: int = 100,
+    verbose: bool = False,
+) -> BenchmarkReport:
+    """Run all benchmarks and return report."""
+    from datetime import datetime
+
+    if displays is None:
+        displays = get_available_displays()
+
+    if effects is None:
+        effects = get_available_effects()
+
+    buffer = get_sample_buffer(80, 24)
+    results = []
+
+    if verbose:
+        print(f"Running benchmarks ({iterations} iterations each)...")
+
+    for name, display_class in displays:
+        if verbose:
+            print(f"Benchmarking display: {name}")
+
+        result = benchmark_display(display_class, buffer, iterations)
+        if result:
+            results.append(result)
+            if verbose:
+                print(f"  {result.fps:.1f} FPS, {result.avg_time_ms:.2f}ms avg")
+
+    if verbose:
+        print()
+
+    for effect_name, effect_class in effects:
+        for display_name, display_class in displays:
+            if display_name == "websocket":
+                continue
+            if verbose:
+                print(f"Benchmarking effect: {effect_name} with {display_name}")
+
+            display = display_class()
+            display.init(80, 24)
+            result = benchmark_effect_with_display(
+                effect_class, display, buffer, iterations
+            )
+            if result:
+                results.append(result)
+                if verbose:
+                    print(f"  {result.fps:.1f} FPS, {result.avg_time_ms:.2f}ms avg")
+
+    summary = generate_summary(results)
+
+    return BenchmarkReport(
+        timestamp=datetime.now().isoformat(),
+        python_version=sys.version,
+        results=results,
+        summary=summary,
+    )
+
+
+def generate_summary(results: list[BenchmarkResult]) -> dict[str, Any]:
+    """Generate summary statistics from results."""
+    by_display: dict[str, list[BenchmarkResult]] = {}
+    by_effect: dict[str, list[BenchmarkResult]] = {}
+
+    for r in results:
+        if r.display not in by_display:
+            by_display[r.display] = []
+        by_display[r.display].append(r)
+
+        if r.effect:
+            if r.effect not in by_effect:
+                by_effect[r.effect] = []
+            by_effect[r.effect].append(r)
+
+    summary = {
+        "by_display": {},
+        "by_effect": {},
+        "overall": {
+            "total_tests": len(results),
+            "displays_tested": len(by_display),
+            "effects_tested": len(by_effect),
+        },
+    }
+
+    for display, res in by_display.items():
+        fps_values = [r.fps for r in res]
+        summary["by_display"][display] = {
+            "avg_fps": float(np.mean(fps_values)),
+            "min_fps": float(np.min(fps_values)),
+            "max_fps": float(np.max(fps_values)),
+            "tests": len(res),
+        }
+
+    for effect, res in by_effect.items():
+        fps_values = [r.fps for r in res]
+        summary["by_effect"][effect] = {
+            "avg_fps": float(np.mean(fps_values)),
+            "min_fps": float(np.min(fps_values)),
+            "max_fps": float(np.max(fps_values)),
+            "tests": len(res),
+        }
+
+    return summary
+
+
+DEFAULT_CACHE_PATH = Path.home() / ".mainline_benchmark_cache.json"
+
+
+def load_baseline(cache_path: Path | None = None) -> dict[str, Any] | None:
+    """Load baseline benchmark results from cache."""
+    path = cache_path or DEFAULT_CACHE_PATH
+    if not path.exists():
+        return None
+    try:
+        with open(path) as f:
+            return json.load(f)
+    except Exception:
+        return None
+
+
+def save_baseline(
+    results: list[BenchmarkResult],
+    cache_path: Path | None = None,
+) -> None:
+    """Save benchmark results as baseline to cache."""
+    path = cache_path or DEFAULT_CACHE_PATH
+    baseline = {
+        "timestamp": datetime.now().isoformat(),
+        "results": {
+            r.name: {
+                "fps": r.fps,
+                "avg_time_ms": r.avg_time_ms,
+                "chars_per_sec": r.chars_per_sec,
+            }
+            for r in results
+        },
+    }
+    path.parent.mkdir(parents=True, exist_ok=True)
+    with open(path, "w") as f:
+        json.dump(baseline, f, indent=2)
+
+
+def compare_with_baseline(
+    results: list[BenchmarkResult],
+    baseline: dict[str, Any],
+    threshold: float = 0.2,
+    verbose: bool = True,
+) -> tuple[bool, list[str]]:
+    """Compare current results with baseline. Returns (pass, messages)."""
+    baseline_results = baseline.get("results", {})
+    failures = []
+    warnings = []
+
+    for r in results:
+        if r.name not in baseline_results:
+            warnings.append(f"New test: {r.name} (no baseline)")
+            continue
+
+        b = baseline_results[r.name]
+        if b["fps"] == 0:
+            continue
+
+        degradation = (b["fps"] - r.fps) / b["fps"]
+        if degradation > threshold:
+            failures.append(
+                f"{r.name}: FPS degraded {degradation * 100:.1f}% "
+                f"(baseline: {b['fps']:.1f}, current: {r.fps:.1f})"
+            )
+        elif verbose:
+            print(f"  {r.name}: {r.fps:.1f} FPS (baseline: {b['fps']:.1f})")
+
+    passed = len(failures) == 0
+    messages = []
+    if failures:
+        messages.extend(failures)
+    if warnings:
+        messages.extend(warnings)
+
+    return passed, messages
+
+
+def run_hook_mode(
+    displays: list[tuple[str, Any]] | None = None,
+    effects: list[tuple[str, Any]] | None = None,
+    iterations: int = 20,
+    threshold: float = 0.2,
+    cache_path: Path | None = None,
+    verbose: bool = False,
+) -> int:
+    """Run in hook mode: compare against baseline, exit 0 on pass, 1 on fail."""
+    baseline = load_baseline(cache_path)
+
+    if baseline is None:
+        print("No baseline found. Run with --baseline to create one.")
+        return 1
+
+    report = run_benchmarks(displays, effects, iterations, verbose)
+
+    passed, messages = compare_with_baseline(
+        report.results, baseline, threshold, verbose
+    )
+
+    print("\n=== Benchmark Hook Results ===")
+    if passed:
+        print("PASSED - No significant performance degradation")
+        return 0
+    else:
+        print("FAILED - Performance degradation detected:")
+        for msg in messages:
+            print(f"  - {msg}")
+        return 1
+
+
+def format_report_text(report: BenchmarkReport) -> str:
+    """Format report as human-readable text."""
+    lines = [
+        "# Mainline Performance Benchmark Report",
+        "",
+        f"Generated: {report.timestamp}",
+        f"Python: {report.python_version}",
+        "",
+        "## Summary",
+        "",
+        f"Total tests: {report.summary['overall']['total_tests']}",
+        f"Displays tested: {report.summary['overall']['displays_tested']}",
+        f"Effects tested: {report.summary['overall']['effects_tested']}",
+        "",
+        "## By Display",
+        "",
+    ]
+
+    for display, stats in report.summary["by_display"].items():
+        lines.append(f"### {display}")
+        lines.append(f"- Avg FPS: {stats['avg_fps']:.1f}")
+        lines.append(f"- Min FPS: {stats['min_fps']:.1f}")
+        lines.append(f"- Max FPS: {stats['max_fps']:.1f}")
+        lines.append(f"- Tests: {stats['tests']}")
+        lines.append("")
+
+    if report.summary["by_effect"]:
+        lines.append("## By Effect")
+        lines.append("")
+
+        for effect, stats in report.summary["by_effect"].items():
+            lines.append(f"### {effect}")
+            lines.append(f"- Avg FPS: {stats['avg_fps']:.1f}")
+            lines.append(f"- Min FPS: {stats['min_fps']:.1f}")
+            lines.append(f"- Max FPS: {stats['max_fps']:.1f}")
+            lines.append(f"- Tests: {stats['tests']}")
+            lines.append("")
+
+    lines.append("## Detailed Results")
+    lines.append("")
+    lines.append("| Display | Effect | FPS | Avg ms | StdDev ms | Min ms | Max ms |")
+    lines.append("|---------|--------|-----|--------|-----------|--------|--------|")
+
+    for r in report.results:
+        effect_col = r.effect if r.effect else "-"
+        lines.append(
+            f"| {r.display} | {effect_col} | {r.fps:.1f} | {r.avg_time_ms:.2f} | "
+            f"{r.std_dev_ms:.2f} | {r.min_ms:.2f} | {r.max_ms:.2f} |"
+        )
+
+    return "\n".join(lines)
+
+
+def format_report_json(report: BenchmarkReport) -> str:
+    """Format report as JSON."""
+    data = {
+        "timestamp": report.timestamp,
+        "python_version": report.python_version,
+        "summary": report.summary,
+        "results": [
+            {
+                "name": r.name,
+                "display": r.display,
+                "effect": r.effect,
+                "iterations": r.iterations,
+                "total_time_ms": r.total_time_ms,
+                "avg_time_ms": r.avg_time_ms,
+                "std_dev_ms": r.std_dev_ms,
+                "min_ms": r.min_ms,
+                "max_ms": r.max_ms,
+                "fps": r.fps,
+                "chars_processed": r.chars_processed,
+                "chars_per_sec": r.chars_per_sec,
+            }
+            for r in report.results
+        ],
+    }
+    return json.dumps(data, indent=2)
+
+
+def main():
+    parser = argparse.ArgumentParser(description="Run mainline benchmarks")
+    parser.add_argument(
+        "--displays",
+        help="Comma-separated list of displays to test (default: all)",
+    )
+    parser.add_argument(
+        "--effects",
+        help="Comma-separated list of effects to test (default: all)",
+    )
+    parser.add_argument(
+        "--iterations",
+        type=int,
+        default=100,
+        help="Number of iterations per test (default: 100)",
+    )
+    parser.add_argument(
+        "--output",
+        help="Output file path (default: stdout)",
+    )
+    parser.add_argument(
+        "--format",
+        choices=["text", "json"],
+        default="text",
+        help="Output format (default: text)",
+    )
+    parser.add_argument(
+        "--verbose",
+        "-v",
+        action="store_true",
+        help="Show progress during benchmarking",
+    )
+    parser.add_argument(
+        "--hook",
+        action="store_true",
+        help="Run in hook mode: compare against baseline, exit 0 pass, 1 fail",
+    )
+    parser.add_argument(
+        "--baseline",
+        action="store_true",
+        help="Save current results as baseline for future hook comparisons",
+    )
+    parser.add_argument(
+        "--threshold",
+        type=float,
+        default=0.2,
+        help="Performance degradation threshold for hook mode (default: 0.2 = 20%%)",
+    )
+    parser.add_argument(
+        "--cache",
+        type=str,
+        default=None,
+        help="Path to baseline cache file (default: ~/.mainline_benchmark_cache.json)",
+    )
+
+    args = parser.parse_args()
+
+    cache_path = Path(args.cache) if args.cache else DEFAULT_CACHE_PATH
+
+    if args.hook:
+        displays = None
+        if args.displays:
+            display_map = dict(get_available_displays())
+            displays = [
+                (name, display_map[name])
+                for name in args.displays.split(",")
+                if name in display_map
+            ]
+
+        effects = None
+        if args.effects:
+            effect_map = dict(get_available_effects())
+            effects = [
+                (name, effect_map[name])
+                for name in args.effects.split(",")
+                if name in effect_map
+            ]
+
+        return run_hook_mode(
+            displays,
+            effects,
+            iterations=args.iterations,
+            threshold=args.threshold,
+            cache_path=cache_path,
+            verbose=args.verbose,
+        )
+
+    displays = None
+    if args.displays:
+        display_map = dict(get_available_displays())
+        displays = [
+            (name, display_map[name])
+            for name in args.displays.split(",")
+            if name in display_map
+        ]
+
+    effects = None
+    if args.effects:
+        effect_map = dict(get_available_effects())
+        effects = [
+            (name, effect_map[name])
+            for name in args.effects.split(",")
+            if name in effect_map
+        ]
+
+    report = run_benchmarks(displays, effects, args.iterations, args.verbose)
+
+    if args.baseline:
+        save_baseline(report.results, cache_path)
+        print(f"Baseline saved to {cache_path}")
+        return 0
+
+    if args.format == "json":
+        output = format_report_json(report)
+    else:
+        output = format_report_text(report)
+
+    if args.output:
+        with open(args.output, "w") as f:
+            f.write(output)
+    else:
+        print(output)
+
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())
--- a/engine/controller.py
+++ b/engine/controller.py
@@ -3,23 +3,48 @@ Stream controller - manages input sources and orchestrates the render stream.
 """

 from engine.config import Config, get_config
+from engine.display import (
+    DisplayRegistry,
+    MultiDisplay,
+    NullDisplay,
+    SixelDisplay,
+    TerminalDisplay,
+    WebSocketDisplay,
+)
 from engine.effects.controller import handle_effects_command
 from engine.eventbus import EventBus
 from engine.events import EventType, StreamEvent
 from engine.mic import MicMonitor
 from engine.ntfy import NtfyPoller
 from engine.scroll import stream
-from engine.websocket_display import WebSocketDisplay


 def _get_display(config: Config):
    """Get the appropriate display based on config."""
-    if config.websocket:
+    DisplayRegistry.initialize()
+    display_mode = config.display.lower()
+
+    displays = []
+
+    if display_mode in ("terminal", "both"):
+        displays.append(TerminalDisplay())
+
+    if display_mode in ("websocket", "both"):
        ws = WebSocketDisplay(host="0.0.0.0", port=config.websocket_port)
        ws.start_server()
        ws.start_http_server()
-        return ws
-    return None
+        displays.append(ws)
+
+    if display_mode == "sixel":
+        displays.append(SixelDisplay())
+
+    if not displays:
+        return NullDisplay()
+
+    if len(displays) == 1:
+        return displays[0]
+
+    return MultiDisplay(displays)


 class StreamController:
--- a/engine/display.py
+++ b/engine/display.py
@@ -1,129 +0,0 @@
-"""
-Display output abstraction - allows swapping output backends.
-
-Protocol:
-    - init(width, height): Initialize display with terminal dimensions
-    - show(buffer): Render buffer (list of strings) to display
-    - clear(): Clear the display
-    - cleanup(): Shutdown display
-"""
-
-import time
-from typing import Protocol
-
-
-class Display(Protocol):
-    """Protocol for display backends."""
-
-    def init(self, width: int, height: int) -> None:
-        """Initialize display with dimensions."""
-        ...
-
-    def show(self, buffer: list[str]) -> None:
-        """Show buffer on display."""
-        ...
-
-    def clear(self) -> None:
-        """Clear display."""
-        ...
-
-    def cleanup(self) -> None:
-        """Shutdown display."""
-        ...
-
-
-def get_monitor():
-    """Get the performance monitor."""
-    try:
-        from engine.effects.performance import get_monitor as _get_monitor
-
-        return _get_monitor()
-    except Exception:
-        return None
-
-
-class TerminalDisplay:
-    """ANSI terminal display backend."""
-
-    def __init__(self):
-        self.width = 80
-        self.height = 24
-
-    def init(self, width: int, height: int) -> None:
-        from engine.terminal import CURSOR_OFF
-
-        self.width = width
-        self.height = height
-        print(CURSOR_OFF, end="", flush=True)
-
-    def show(self, buffer: list[str]) -> None:
-        import sys
-
-        t0 = time.perf_counter()
-        sys.stdout.buffer.write("".join(buffer).encode())
-        sys.stdout.flush()
-        elapsed_ms = (time.perf_counter() - t0) * 1000
-
-        monitor = get_monitor()
-        if monitor:
-            chars_in = sum(len(line) for line in buffer)
-            monitor.record_effect("terminal_display", elapsed_ms, chars_in, chars_in)
-
-    def clear(self) -> None:
-        from engine.terminal import CLR
-
-        print(CLR, end="", flush=True)
-
-    def cleanup(self) -> None:
-        from engine.terminal import CURSOR_ON
-
-        print(CURSOR_ON, end="", flush=True)
-
-
-class NullDisplay:
-    """Headless/null display - discards all output."""
-
-    def init(self, width: int, height: int) -> None:
-        self.width = width
-        self.height = height
-
-    def show(self, buffer: list[str]) -> None:
-        monitor = get_monitor()
-        if monitor:
-            t0 = time.perf_counter()
-            chars_in = sum(len(line) for line in buffer)
-            elapsed_ms = (time.perf_counter() - t0) * 1000
-            monitor.record_effect("null_display", elapsed_ms, chars_in, chars_in)
-
-    def clear(self) -> None:
-        pass
-
-    def cleanup(self) -> None:
-        pass
-
-
-class MultiDisplay:
-    """Display that forwards to multiple displays."""
-
-    def __init__(self, displays: list[Display]):
-        self.displays = displays
-        self.width = 80
-        self.height = 24
-
-    def init(self, width: int, height: int) -> None:
-        self.width = width
-        self.height = height
-        for d in self.displays:
-            d.init(width, height)
-
-    def show(self, buffer: list[str]) -> None:
-        for d in self.displays:
-            d.show(buffer)
-
-    def clear(self) -> None:
-        for d in self.displays:
-            d.clear()
-
-    def cleanup(self) -> None:
-        for d in self.displays:
-            d.cleanup()
--- a/engine/display/init.py
+++ b/engine/display/init.py
@@ -0,0 +1,102 @@
+"""
+Display backend system with registry pattern.
+
+Allows swapping output backends via the Display protocol.
+Supports auto-discovery of display backends.
+"""
+
+from typing import Protocol
+
+from engine.display.backends.multi import MultiDisplay
+from engine.display.backends.null import NullDisplay
+from engine.display.backends.sixel import SixelDisplay
+from engine.display.backends.terminal import TerminalDisplay
+from engine.display.backends.websocket import WebSocketDisplay
+
+
+class Display(Protocol):
+    """Protocol for display backends."""
+
+    width: int
+    height: int
+
+    def init(self, width: int, height: int) -> None:
+        """Initialize display with dimensions."""
+        ...
+
+    def show(self, buffer: list[str]) -> None:
+        """Show buffer on display."""
+        ...
+
+    def clear(self) -> None:
+        """Clear display."""
+        ...
+
+    def cleanup(self) -> None:
+        """Shutdown display."""
+        ...
+
+
+class DisplayRegistry:
+    """Registry for display backends with auto-discovery."""
+
+    _backends: dict[str, type[Display]] = {}
+    _initialized = False
+
+    @classmethod
+    def register(cls, name: str, backend_class: type[Display]) -> None:
+        """Register a display backend."""
+        cls._backends[name.lower()] = backend_class
+
+    @classmethod
+    def get(cls, name: str) -> type[Display] | None:
+        """Get a display backend class by name."""
+        return cls._backends.get(name.lower())
+
+    @classmethod
+    def list_backends(cls) -> list[str]:
+        """List all available display backend names."""
+        return list(cls._backends.keys())
+
+    @classmethod
+    def create(cls, name: str, **kwargs) -> Display | None:
+        """Create a display instance by name."""
+        backend_class = cls.get(name)
+        if backend_class:
+            return backend_class(**kwargs)
+        return None
+
+    @classmethod
+    def initialize(cls) -> None:
+        """Initialize and register all built-in backends."""
+        if cls._initialized:
+            return
+
+        cls.register("terminal", TerminalDisplay)
+        cls.register("null", NullDisplay)
+        cls.register("websocket", WebSocketDisplay)
+        cls.register("sixel", SixelDisplay)
+
+        cls._initialized = True
+
+
+def get_monitor():
+    """Get the performance monitor."""
+    try:
+        from engine.effects.performance import get_monitor as _get_monitor
+
+        return _get_monitor()
+    except Exception:
+        return None
+
+
+__all__ = [
+    "Display",
+    "DisplayRegistry",
+    "get_monitor",
+    "TerminalDisplay",
+    "NullDisplay",
+    "WebSocketDisplay",
+    "SixelDisplay",
+    "MultiDisplay",
+]
--- a/engine/display/backends/multi.py
+++ b/engine/display/backends/multi.py
@@ -0,0 +1,33 @@
+"""
+Multi display backend - forwards to multiple displays.
+"""
+
+
+class MultiDisplay:
+    """Display that forwards to multiple displays."""
+
+    width: int = 80
+    height: int = 24
+
+    def __init__(self, displays: list):
+        self.displays = displays
+        self.width = 80
+        self.height = 24
+
+    def init(self, width: int, height: int) -> None:
+        self.width = width
+        self.height = height
+        for d in self.displays:
+            d.init(width, height)
+
+    def show(self, buffer: list[str]) -> None:
+        for d in self.displays:
+            d.show(buffer)
+
+    def clear(self) -> None:
+        for d in self.displays:
+            d.clear()
+
+    def cleanup(self) -> None:
+        for d in self.displays:
+            d.cleanup()
--- a/engine/display/backends/null.py
+++ b/engine/display/backends/null.py
@@ -0,0 +1,32 @@
+"""
+Null/headless display backend.
+"""
+
+import time
+
+
+class NullDisplay:
+    """Headless/null display - discards all output."""
+
+    width: int = 80
+    height: int = 24
+
+    def init(self, width: int, height: int) -> None:
+        self.width = width
+        self.height = height
+
+    def show(self, buffer: list[str]) -> None:
+        from engine.display import get_monitor
+
+        monitor = get_monitor()
+        if monitor:
+            t0 = time.perf_counter()
+            chars_in = sum(len(line) for line in buffer)
+            elapsed_ms = (time.perf_counter() - t0) * 1000
+            monitor.record_effect("null_display", elapsed_ms, chars_in, chars_in)
+
+    def clear(self) -> None:
+        pass
+
+    def cleanup(self) -> None:
+        pass
--- a/engine/display/backends/sixel.py
+++ b/engine/display/backends/sixel.py
@@ -0,0 +1,269 @@
+"""
+Sixel graphics display backend - renders to sixel graphics in terminal.
+"""
+
+import time
+
+
+def _parse_ansi(
+    text: str,
+) -> list[tuple[str, tuple[int, int, int], tuple[int, int, int], bool]]:
+    """Parse ANSI text into tokens with fg/bg colors.
+
+    Returns list of (text, fg_rgb, bg_rgb, bold).
+    """
+    tokens = []
+    current_text = ""
+    fg = (204, 204, 204)
+    bg = (0, 0, 0)
+    bold = False
+    i = 0
+
+    ANSI_COLORS = {
+        0: (0, 0, 0),
+        1: (205, 49, 49),
+        2: (13, 188, 121),
+        3: (229, 229, 16),
+        4: (36, 114, 200),
+        5: (188, 63, 188),
+        6: (17, 168, 205),
+        7: (229, 229, 229),
+        8: (102, 102, 102),
+        9: (241, 76, 76),
+        10: (35, 209, 139),
+        11: (245, 245, 67),
+        12: (59, 142, 234),
+        13: (214, 112, 214),
+        14: (41, 184, 219),
+        15: (255, 255, 255),
+    }
+
+    while i < len(text):
+        char = text[i]
+
+        if char == "\x1b" and i + 1 < len(text) and text[i + 1] == "[":
+            if current_text:
+                tokens.append((current_text, fg, bg, bold))
+                current_text = ""
+
+            i += 2
+            code = ""
+            while i < len(text):
+                c = text[i]
+                if c.isalpha():
+                    break
+                code += c
+                i += 1
+
+            if code:
+                codes = code.split(";")
+                for c in codes:
+                    try:
+                        n = int(c) if c else 0
+                    except ValueError:
+                        continue
+
+                    if n == 0:
+                        fg = (204, 204, 204)
+                        bg = (0, 0, 0)
+                        bold = False
+                    elif n == 1:
+                        bold = True
+                    elif n == 22:
+                        bold = False
+                    elif n == 39:
+                        fg = (204, 204, 204)
+                    elif n == 49:
+                        bg = (0, 0, 0)
+                    elif 30 <= n <= 37:
+                        fg = ANSI_COLORS.get(n - 30 + (8 if bold else 0), fg)
+                    elif 40 <= n <= 47:
+                        bg = ANSI_COLORS.get(n - 40, bg)
+                    elif 90 <= n <= 97:
+                        fg = ANSI_COLORS.get(n - 90 + 8, fg)
+                    elif 100 <= n <= 107:
+                        bg = ANSI_COLORS.get(n - 100 + 8, bg)
+                    elif 1 <= n <= 256:
+                        if n < 16:
+                            fg = ANSI_COLORS.get(n, fg)
+                        elif n < 232:
+                            c = n - 16
+                            r = (c // 36) * 51
+                            g = ((c % 36) // 6) * 51
+                            b = (c % 6) * 51
+                            fg = (r, g, b)
+                        else:
+                            gray = (n - 232) * 10 + 8
+                            fg = (gray, gray, gray)
+        else:
+            current_text += char
+            i += 1
+
+    if current_text:
+        tokens.append((current_text, fg, bg, bold))
+
+    return tokens if tokens else [("", fg, bg, bold)]
+
+
+def _encode_sixel(image) -> str:
+    """Encode a PIL Image to sixel format (pure Python)."""
+    img = image.convert("RGBA")
+    width, height = img.size
+    pixels = img.load()
+
+    palette = []
+    pixel_palette_idx = {}
+
+    def get_color_idx(r, g, b, a):
+        if a < 128:
+            return -1
+        key = (r // 32, g // 32, b // 32)
+        if key not in pixel_palette_idx:
+            idx = len(palette)
+            if idx < 256:
+                palette.append((r, g, b))
+                pixel_palette_idx[key] = idx
+        return pixel_palette_idx.get(key, 0)
+
+    for y in range(height):
+        for x in range(width):
+            r, g, b, a = pixels[x, y]
+            get_color_idx(r, g, b, a)
+
+    if not palette:
+        return ""
+
+    if len(palette) == 1:
+        palette = [palette[0], (0, 0, 0)]
+
+    sixel_data = []
+    sixel_data.append(
+        f'"{"".join(f"#{i};2;{r};{g};{b}" for i, (r, g, b) in enumerate(palette))}'
+    )
+
+    for x in range(width):
+        col_data = []
+        for y in range(0, height, 6):
+            bits = 0
+            color_idx = -1
+            for dy in range(6):
+                if y + dy < height:
+                    r, g, b, a = pixels[x, y + dy]
+                    if a >= 128:
+                        bits |= 1 << dy
+                        idx = get_color_idx(r, g, b, a)
+                        if color_idx == -1:
+                            color_idx = idx
+                        elif color_idx != idx:
+                            color_idx = -2
+
+            if color_idx >= 0:
+                col_data.append(
+                    chr(63 + color_idx) + chr(63 + bits)
+                    if bits
+                    else chr(63 + color_idx) + "?"
+                )
+            elif color_idx == -2:
+                pass
+
+        if col_data:
+            sixel_data.append("".join(col_data) + "$")
+        else:
+            sixel_data.append("-" if x < width - 1 else "$")
+
+    sixel_data.append("\x1b\\")
+
+    return "\x1bPq" + "".join(sixel_data)
+
+
+class SixelDisplay:
+    """Sixel graphics display backend - renders to sixel graphics in terminal."""
+
+    width: int = 80
+    height: int = 24
+
+    def __init__(self, cell_width: int = 9, cell_height: int = 16):
+        self.width = 80
+        self.height = 24
+        self.cell_width = cell_width
+        self.cell_height = cell_height
+        self._initialized = False
+
+    def init(self, width: int, height: int) -> None:
+        self.width = width
+        self.height = height
+        self._initialized = True
+
+    def show(self, buffer: list[str]) -> None:
+        import sys
+
+        t0 = time.perf_counter()
+
+        img_width = self.width * self.cell_width
+        img_height = self.height * self.cell_height
+
+        try:
+            from PIL import Image, ImageDraw, ImageFont
+        except ImportError:
+            return
+
+        img = Image.new("RGBA", (img_width, img_height), (0, 0, 0, 255))
+        draw = ImageDraw.Draw(img)
+
+        try:
+            font = ImageFont.truetype(
+                "/usr/share/fonts/truetype/dejavu/DejaVuSansMono.ttf",
+                self.cell_height - 2,
+            )
+        except Exception:
+            try:
+                font = ImageFont.load_default()
+            except Exception:
+                font = None
+
+        for row_idx, line in enumerate(buffer[: self.height]):
+            if row_idx >= self.height:
+                break
+
+            tokens = _parse_ansi(line)
+            x_pos = 0
+            y_pos = row_idx * self.cell_height
+
+            for text, fg, bg, bold in tokens:
+                if not text:
+                    continue
+
+                if bg != (0, 0, 0):
+                    bbox = draw.textbbox((x_pos, y_pos), text, font=font)
+                    draw.rectangle(bbox, fill=(*bg, 255))
+
+                if bold and font:
+                    draw.text((x_pos - 1, y_pos - 1), text, fill=(*fg, 255), font=font)
+
+                draw.text((x_pos, y_pos), text, fill=(*fg, 255), font=font)
+
+                if font:
+                    x_pos += draw.textlength(text, font=font)
+
+        sixel = _encode_sixel(img)
+
+        sys.stdout.buffer.write(sixel.encode("utf-8"))
+        sys.stdout.flush()
+
+        elapsed_ms = (time.perf_counter() - t0) * 1000
+
+        from engine.display import get_monitor
+
+        monitor = get_monitor()
+        if monitor:
+            chars_in = sum(len(line) for line in buffer)
+            monitor.record_effect("sixel_display", elapsed_ms, chars_in, chars_in)
+
+    def clear(self) -> None:
+        import sys
+
+        sys.stdout.buffer.write(b"\x1b[2J\x1b[H")
+        sys.stdout.flush()
+
+    def cleanup(self) -> None:
+        pass
--- a/engine/display/backends/terminal.py
+++ b/engine/display/backends/terminal.py
@@ -0,0 +1,48 @@
+"""
+ANSI terminal display backend.
+"""
+
+import time
+
+
+class TerminalDisplay:
+    """ANSI terminal display backend."""
+
+    width: int = 80
+    height: int = 24
+
+    def __init__(self):
+        self.width = 80
+        self.height = 24
+
+    def init(self, width: int, height: int) -> None:
+        from engine.terminal import CURSOR_OFF
+
+        self.width = width
+        self.height = height
+        print(CURSOR_OFF, end="", flush=True)
+
+    def show(self, buffer: list[str]) -> None:
+        import sys
+
+        t0 = time.perf_counter()
+        sys.stdout.buffer.write("".join(buffer).encode())
+        sys.stdout.flush()
+        elapsed_ms = (time.perf_counter() - t0) * 1000
+
+        from engine.display import get_monitor
+
+        monitor = get_monitor()
+        if monitor:
+            chars_in = sum(len(line) for line in buffer)
+            monitor.record_effect("terminal_display", elapsed_ms, chars_in, chars_in)
+
+    def clear(self) -> None:
+        from engine.terminal import CLR
+
+        print(CLR, end="", flush=True)
+
+    def cleanup(self) -> None:
+        from engine.terminal import CURSOR_ON
+
+        print(CURSOR_ON, end="", flush=True)
--- a/engine/display/backends/websocket.py
+++ b/engine/display/backends/websocket.py
@@ -1,11 +1,5 @@
 """
-WebSocket display server - broadcasts frame buffer to connected web clients.
-
-Usage:
-    ws_display = WebSocketDisplay(host="0.0.0.0", port=8765)
-    ws_display.init(80, 24)
-    ws_display.show(["line1", "line2", ...])
-    ws_display.cleanup()
+WebSocket display backend - broadcasts frame buffer to connected web clients.
 """

 import asyncio
@@ -23,6 +17,9 @@ except ImportError:
 class Display(Protocol):
    """Protocol for display backends."""

+    width: int
+    height: int
+
    def init(self, width: int, height: int) -> None:
        """Initialize display with dimensions."""
        ...
@@ -53,6 +50,9 @@ def get_monitor():
 class WebSocketDisplay:
    """WebSocket display backend - broadcasts to HTML Canvas clients."""

+    width: int = 80
+    height: int = 24
+
    def __init__(
        self,
        host: str = "0.0.0.0",
@@ -177,7 +177,9 @@ class WebSocketDisplay:
        import os
        from http.server import HTTPServer, SimpleHTTPRequestHandler

-        client_dir = os.path.join(os.path.dirname(os.path.dirname(__file__)), "client")
+        client_dir = os.path.join(
+            os.path.dirname(os.path.dirname(os.path.dirname(__file__))), "client"
+        )

        class Handler(SimpleHTTPRequestHandler):
            def __init__(self, *args, **kwargs):
--- a/hk.pkl
+++ b/hk.pkl
@@ -22,6 +22,9 @@ hooks {
                prefix = "uv run"
                check = "ruff check engine/ tests/"
            }
+            ["benchmark"] {
+                check = "uv run python -m engine.benchmark --hook --displays null --iterations 20"
+            }
        }
    }
 }
--- a/mise.toml
+++ b/mise.toml
@@ -33,6 +33,7 @@ run-poetry = "uv run mainline.py --poetry"
 run-firehose = "uv run mainline.py --firehose"

 run-websocket = { run = "uv run mainline.py --display websocket", depends = ["sync-all"] }
+run-sixel = { run = "uv run mainline.py --display sixel", depends = ["sync-all"] }
 run-both = { run = "uv run mainline.py --display both", depends = ["sync-all"] }
 run-client = { run = "mise run run-both & sleep 2 && $(open http://localhost:8766 2>/dev/null || xdg-open http://localhost:8766 2>/dev/null || echo 'Open http://localhost:8766 manually'); wait", depends = ["sync-all"] }

@@ -43,6 +44,14 @@ run-client = { run = "mise run run-both & sleep 2 && $(open http://localhost:876
 cmd = "uv run cmdline.py"
 cmd-stats = { run = "uv run cmdline.py -w \"/effects stats\"", depends = ["sync-all"] }

+# =====================
+# Benchmark
+# =====================
+
+benchmark = { run = "uv run python -m engine.benchmark", depends = ["sync-all"] }
+benchmark-json = { run = "uv run python -m engine.benchmark --format json --output benchmark.json", depends = ["sync-all"] }
+benchmark-report = { run = "uv run python -m engine.benchmark --output BENCHMARK.md", depends = ["sync-all"] }
+
 # Initialize ntfy topics (warm up before first use)
 topics-init = "curl -s -d 'init' https://ntfy.sh/klubhaus_terminal_mainline_cc_cmd > /dev/null && curl -s -d 'init' https://ntfy.sh/klubhaus_terminal_mainline_cc_resp > /dev/null && curl -s -d 'init' https://ntfy.sh/klubhaus_terminal_mainline > /dev/null"

--- a/pyproject.toml
+++ b/pyproject.toml
@@ -33,6 +33,9 @@ mic = [
 websocket = [
    "websockets>=12.0",
 ]
+sixel = [
+    "pysixel>=0.1.0",
+]
 browser = [
    "playwright>=1.40.0",
 ]
--- a/tests/test_websocket.py
+++ b/tests/test_websocket.py
@@ -1,12 +1,12 @@
 """
-Tests for engine.websocket_display module.
+Tests for engine.display.backends.websocket module.
 """

 from unittest.mock import MagicMock, patch

 import pytest

-from engine.websocket_display import WebSocketDisplay
+from engine.display.backends.websocket import WebSocketDisplay


 class TestWebSocketDisplayImport:
@@ -14,9 +14,9 @@ class TestWebSocketDisplayImport:

    def test_import_does_not_error(self):
        """Module imports without error."""
-        from engine import websocket_display
+        from engine.display import backends

-        assert websocket_display is not None
+        assert backends is not None


 class TestWebSocketDisplayInit:
@@ -24,7 +24,7 @@ class TestWebSocketDisplayInit:

    def test_default_init(self):
        """Default initialization sets correct defaults."""
-        with patch("engine.websocket_display.websockets", None):
+        with patch("engine.display.backends.websocket.websockets", None):
            display = WebSocketDisplay()
            assert display.host == "0.0.0.0"
            assert display.port == 8765
@@ -34,7 +34,7 @@ class TestWebSocketDisplayInit:

    def test_custom_init(self):
        """Custom initialization uses provided values."""
-        with patch("engine.websocket_display.websockets", None):
+        with patch("engine.display.backends.websocket.websockets", None):
            display = WebSocketDisplay(host="localhost", port=9000, http_port=9001)
            assert display.host == "localhost"
            assert display.port == 9000
@@ -60,7 +60,7 @@ class TestWebSocketDisplayProtocol:

    def test_websocket_display_is_display(self):
        """WebSocketDisplay satisfies Display protocol."""
-        with patch("engine.websocket_display.websockets", MagicMock()):
+        with patch("engine.display.backends.websocket.websockets", MagicMock()):
            display = WebSocketDisplay()
            assert hasattr(display, "init")
            assert hasattr(display, "show")
@@ -73,7 +73,7 @@ class TestWebSocketDisplayMethods:

    def test_init_stores_dimensions(self):
        """init stores terminal dimensions."""
-        with patch("engine.websocket_display.websockets", MagicMock()):
+        with patch("engine.display.backends.websocket.websockets", MagicMock()):
            display = WebSocketDisplay()
            display.init(100, 40)
            assert display.width == 100
@@ -81,31 +81,31 @@ class TestWebSocketDisplayMethods:

    def test_client_count_initially_zero(self):
        """client_count returns 0 when no clients connected."""
-        with patch("engine.websocket_display.websockets", MagicMock()):
+        with patch("engine.display.backends.websocket.websockets", MagicMock()):
            display = WebSocketDisplay()
            assert display.client_count() == 0

    def test_get_ws_port(self):
        """get_ws_port returns configured port."""
-        with patch("engine.websocket_display.websockets", MagicMock()):
+        with patch("engine.display.backends.websocket.websockets", MagicMock()):
            display = WebSocketDisplay(port=9000)
            assert display.get_ws_port() == 9000

    def test_get_http_port(self):
        """get_http_port returns configured port."""
-        with patch("engine.websocket_display.websockets", MagicMock()):
+        with patch("engine.display.backends.websocket.websockets", MagicMock()):
            display = WebSocketDisplay(http_port=9001)
            assert display.get_http_port() == 9001

    def test_frame_delay_defaults_to_zero(self):
        """get_frame_delay returns 0 by default."""
-        with patch("engine.websocket_display.websockets", MagicMock()):
+        with patch("engine.display.backends.websocket.websockets", MagicMock()):
            display = WebSocketDisplay()
            assert display.get_frame_delay() == 0.0

    def test_set_frame_delay(self):
        """set_frame_delay stores the value."""
-        with patch("engine.websocket_display.websockets", MagicMock()):
+        with patch("engine.display.backends.websocket.websockets", MagicMock()):
            display = WebSocketDisplay()
            display.set_frame_delay(0.05)
            assert display.get_frame_delay() == 0.05
@@ -116,7 +116,7 @@ class TestWebSocketDisplayCallbacks:

    def test_set_client_connected_callback(self):
        """set_client_connected_callback stores callback."""
-        with patch("engine.websocket_display.websockets", MagicMock()):
+        with patch("engine.display.backends.websocket.websockets", MagicMock()):
            display = WebSocketDisplay()
            callback = MagicMock()
            display.set_client_connected_callback(callback)
@@ -124,7 +124,7 @@ class TestWebSocketDisplayCallbacks:

    def test_set_client_disconnected_callback(self):
        """set_client_disconnected_callback stores callback."""
-        with patch("engine.websocket_display.websockets", MagicMock()):
+        with patch("engine.display.backends.websocket.websockets", MagicMock()):
            display = WebSocketDisplay()
            callback = MagicMock()
            display.set_client_disconnected_callback(callback)