docs: Reorganize documentation with tiered hierarchy for context optimization

Major documentation reorganization to reduce AI agent context size by ~58%
and establish sustainable maintenance practices.

## File Moves (Root → doc/)
- Move COMPLETED.md (new), HANDOFF*.md, *_ANALYSIS.md, *_SUMMARY.md to doc/
- Keep only 5 essential files in root: CLAUDE.md, GEMINI.md, PROJECT_CONTEXT.md, TODO.md, README.md
- Result: Clean root directory with clear project essentials

## New Documentation
- doc/CONTEXT_MAINTENANCE.md: Comprehensive guide for keeping context clean
  - 4-tier hierarchy (Critical/Technical/Design/Archive)
  - Maintenance schedules (after milestones, monthly, on-demand)
  - Size targets, red flags, workflows
  - Monthly checklist template

- doc/COMPLETED.md: Historical archive of completed milestones
  - Moved "Recently Completed" sections from TODO.md and PROJECT_CONTEXT.md
  - Detailed completion history (February 4-7, 2026)
  - Frees up ~200 lines from active context

## Agent Config Updates
- CLAUDE.md: Restructured with 4-tier hierarchy
  - Tier 1: Critical (always loaded) - 3 files
  - Tier 2: Technical (always loaded) - 3 files
  - Tier 3: Design (on-demand) - 9 files
  - Tier 4: Archive (rarely) - 10 files
  - Clear usage instructions for on-demand loading

- GEMINI.md: Same tier structure + Gemini-specific state snapshot
  - Consistent with CLAUDE.md hierarchy
  - Preserved agent-specific context

## Content Optimization
- PROJECT_CONTEXT.md: Removed verbose milestones (~160 lines)
  - Replaced with concise "Current Status" summary
  - Points to COMPLETED.md for history

- TODO.md: Removed Task #51 detailed plan (~200 lines)
  - Marked Task #51 as completed
  - Kept only active/next tasks

## Impact
- Context size: 70K → 29K tokens (58% reduction)
- Root directory: 15 → 5 files (67% cleaner)
- Tier 1-2 files: 7,329 words (well under 10K target)
- Documented maintenance process for sustainability

## Files Changed
Modified: CLAUDE.md, GEMINI.md, PROJECT_CONTEXT.md, TODO.md
New: doc/COMPLETED.md, doc/CONTEXT_MAINTENANCE.md
Moved: 10 technical docs from root to doc/

15 files changed, 1029 insertions, 508 deletions

diff --git a/CLAUDE.md b/CLAUDE.md
index cdc9e0e..f8dcd05 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -1,27 +1,65 @@
-# Project Context
+# ============================================
+# TIER 1: CRITICAL CONTEXT (Always Loaded)
+# ============================================
 @PROJECT_CONTEXT.md
 @TODO.md
-
-# helping and quick-start tutorials
-
 @README.md
-@doc/HOWTO.md
 
-# design docs
-@doc/ASSET_SYSTEM.md
-@doc/BUILD.md
-@doc/FETCH_DEPS.md
-@doc/3D.md
-@doc/SPEC_EDITOR.md
-@doc/TRACKER.md
-@doc/PROCEDURAL.md
-@doc/ANALYSIS_VARIABLE_TEMPO.md
-@doc/ANALYSIS_VARIABLE_TEMPO_V2.md
-
-# coding guidelines
+# ============================================
+# TIER 2: TECHNICAL REFERENCE (Always Loaded)
+# ============================================
+@doc/HOWTO.md
 @doc/CONTRIBUTING.md
 @doc/AI_RULES.md
 
+# ============================================
+# TIER 3: DESIGN DOCS (Load On-Demand)
+# ============================================
+# Load these only when working on specific subsystems:
+#
+# Audio & Tracker:
+#   @doc/SPEC_EDITOR.md - Spectral editor design
+#   @doc/TRACKER.md - Audio tracker system
+#   @doc/ANALYSIS_VARIABLE_TEMPO_V2.md - Variable tempo analysis
+#
+# 3D & Graphics:
+#   @doc/3D.md - 3D rendering architecture
+#   @doc/PROCEDURAL.md - Procedural generation
+#
+# Build & Assets:
+#   @doc/ASSET_SYSTEM.md - Asset pipeline details
+#   @doc/BUILD.md - Build system details
+#   @doc/FETCH_DEPS.md - Dependency management
+#
+# Testing & Tools:
+#   @doc/test_demo_README.md - test_demo tool documentation
+
+# ============================================
+# TIER 4: HISTORICAL ARCHIVE (Load Rarely)
+# ============================================
+# Load these only for historical context or debugging:
+#
+# Completion History:
+#   @doc/COMPLETED.md - Detailed milestone history
+#
+# Technical Investigations:
+#   @doc/GPU_EFFECTS_TEST_ANALYSIS.md
+#   @doc/PLATFORM_ANALYSIS.md
+#   @doc/PLATFORM_SIDE_QUEST_SUMMARY.md
+#   @doc/PEAK_FIX_SUMMARY.md
+#
+# Agent Handoffs:
+#   @doc/HANDOFF_CLAUDE.md
+#   @doc/HANDOFF.md
+#   @doc/HANDOFF_2026-02-04.md
+#
+# Task Tracking:
+#   @doc/TASKS_SUMMARY.md
+
+# ============================================
+# PROJECT RULES (IMPORTANT)
+# ============================================
+
 IMPORTANT:
 - Follow all rules in doc/AI_RULES.md
 - This repository is shared with Gemini-CLI
@@ -29,3 +67,9 @@ IMPORTANT:
 - Work only on tasks explicitly requested by the user
 - Do NOT modify files outside the current scope
 - Do NOT perform refactors or cleanups unless explicitly asked
+
+# Context Maintenance:
+- See @doc/CONTEXT_MAINTENANCE.md for keeping context clean
+- Archive completed work to doc/COMPLETED.md regularly
+- Keep PROJECT_CONTEXT.md focused on current status
+- Keep TODO.md focused on active/next tasks only
diff --git a/GEMINI.md b/GEMINI.md
index b828621..c57c12d 100644
--- a/GEMINI.md
+++ b/GEMINI.md
@@ -1,34 +1,82 @@
-# Project Context
+# ============================================
+# TIER 1: CRITICAL CONTEXT (Always Loaded)
+# ============================================
 @PROJECT_CONTEXT.md
 @TODO.md
-
-# helping and quick-start tutorials
 @README.md
-@doc/HOWTO.md
 
-# design docs
-@doc/ASSET_SYSTEM.md
-@doc/BUILD.md
-@doc/FETCH_DEPS.md
-@doc/3D.md
-@doc/SPEC_EDITOR.md
-@doc/TRACKER.md
-@doc/PROCEDURAL.md
-@doc/ANALYSIS_VARIABLE_TEMPO.md
-@doc/ANALYSIS_VARIABLE_TEMPO_V2.md
-
-# coding guidelines
+# ============================================
+# TIER 2: TECHNICAL REFERENCE (Always Loaded)
+# ============================================
+@doc/HOWTO.md
 @doc/CONTRIBUTING.md
 @doc/AI_RULES.md
 
-# AI rules
+# ============================================
+# TIER 3: DESIGN DOCS (Load On-Demand)
+# ============================================
+# Load these only when working on specific subsystems:
+#
+# Audio & Tracker:
+#   @doc/SPEC_EDITOR.md - Spectral editor design
+#   @doc/TRACKER.md - Audio tracker system
+#   @doc/ANALYSIS_VARIABLE_TEMPO_V2.md - Variable tempo analysis
+#
+# 3D & Graphics:
+#   @doc/3D.md - 3D rendering architecture
+#   @doc/PROCEDURAL.md - Procedural generation
+#
+# Build & Assets:
+#   @doc/ASSET_SYSTEM.md - Asset pipeline details
+#   @doc/BUILD.md - Build system details
+#   @doc/FETCH_DEPS.md - Dependency management
+#
+# Testing & Tools:
+#   @doc/test_demo_README.md - test_demo tool documentation
+
+# ============================================
+# TIER 4: HISTORICAL ARCHIVE (Load Rarely)
+# ============================================
+# Load these only for historical context or debugging:
+#
+# Completion History:
+#   @doc/COMPLETED.md - Detailed milestone history
+#
+# Technical Investigations:
+#   @doc/GPU_EFFECTS_TEST_ANALYSIS.md
+#   @doc/PLATFORM_ANALYSIS.md
+#   @doc/PLATFORM_SIDE_QUEST_SUMMARY.md
+#   @doc/PEAK_FIX_SUMMARY.md
+#
+# Agent Handoffs:
+#   @doc/HANDOFF_CLAUDE.md
+#   @doc/HANDOFF.md
+#   @doc/HANDOFF_2026-02-04.md
+#
+# Task Tracking:
+#   @doc/TASKS_SUMMARY.md
+
+# ============================================
+# PROJECT RULES (IMPORTANT)
+# ============================================
+
+IMPORTANT:
+- Follow all rules in doc/AI_RULES.md
+- This repository is shared with Claude-Code
+- You are working in turns with another AI agent
+- Work only on tasks explicitly requested by the user
+- Do NOT modify files outside the current scope
+- Do NOT perform refactors or cleanups unless explicitly asked
 
-follow rules in AI_RULES.md !
-You are working in a repo shared with Claude-Code.
-Work only on the tasks explicitly requested.
-Do not modify files outside the current scope.
+# Context Maintenance:
+- See @doc/CONTEXT_MAINTENANCE.md for keeping context clean
+- Archive completed work to doc/COMPLETED.md regularly
+- Keep PROJECT_CONTEXT.md focused on current status
+- Keep TODO.md focused on active/next tasks only
 
-# Current State Snapshot
+# ============================================
+# CURRENT STATE SNAPSHOT (Gemini-Specific)
+# ============================================
 <state_snapshot>
     <overall_goal>
         Evolve the demo into a robust 3D engine with dynamic physics, BVH acceleration, and a modular shader architecture, targeting a 64k binary limit.
diff --git a/PROJECT_CONTEXT.md b/PROJECT_CONTEXT.md
index 8f71ebf..273acd7 100644
--- a/PROJECT_CONTEXT.md
+++ b/PROJECT_CONTEXT.md
@@ -30,96 +30,13 @@ Style:
 ---
 ## Project Roadmap
 
-### Recently Completed
+**Note:** For detailed history of recently completed milestones, see `COMPLETED.md`.
 
-#### Milestone: Audio Peak Measurement & Test Coverage (February 7, 2026) ✅
-- **Real-Time Audio Peak Fix**: Resolved critical audio-visual synchronization bug where visual effects triggered ~400ms before audio was heard. Root cause: peak measurement occurred at ring buffer write time (synth_render) instead of playback time (audio callback). Solution: Added `get_realtime_peak()` to AudioBackend interface, implemented real-time peak tracking in MiniaudioBackend audio_callback using exponential averaging (instant attack, 0.7 decay rate for 1-second fade). Updated main.cc and test_demo.cc to use `audio_get_realtime_peak()` API. Result: Perfect audio-visual synchronization, visual effects trigger precisely when audio is heard.
-
-- **Peak Decay Optimization**: Fixed "test_demo just flashing" issue caused by slow decay rate. Old: 0.95 decay = 5.76 second fade (screen stayed white constantly). New: 0.7 decay = 1.15 second fade (proper flash with smooth decay). Mathematical verification: At 128ms audio callbacks, 0.7^9 ≈ 0.1 after ~1 second. Result: Responsive visual effects with proper attack/decay curves.
-
-- **SilentBackend Test Infrastructure**: Created test-only audio backend for comprehensive audio.cc testing without hardware. Features: controllable peak for edge cases, tracks frames rendered and voice triggers, pure inspection (no audio output). Created 7 comprehensive tests covering lifecycle, peak control, frame/voice tracking, playback time, buffer management, audio_update safety. Result: Significantly improved audio.cc test coverage (audio_get_playback_time, audio_render_ahead, audio_update all tested).
-
-- **Backend Reorganization**: Moved all audio backend implementations to src/audio/backend/ subdirectory for cleaner code organization. Moved: miniaudio_backend, mock_audio_backend, wav_dump_backend, jittered_audio_backend, silent_backend (5 backends total). Updated all #include paths in backend files and tests, updated CMakeLists.txt paths. Kept audio_backend.h (interface) in src/audio/. Result: Improved maintainability, zero functionality regressions, all 28 tests pass.
-
-- **Dead Code Removal**: Removed unused `register_spec_asset()` function from audio.{h,cc}. Function was never called anywhere in codebase. Removed declaration and implementation (lines 43-58). Result: Cleaner codebase, reduced maintenance burden.
-
-- **Test Coverage Achievement**: All 28 tests passing (100% success rate). New SilentBackend tests provide comprehensive coverage for audio subsystem. No regressions in existing tests. Build clean with no errors or warnings. Files modified: 15 changed, 3 new files, 8 files reorganized. Commit a6a7bf0 ready for push.
-
-#### Milestone: test_demo - Audio/Visual Sync Debug Tool (February 7, 2026) 🎯
-- **Standalone Debug Tool**: Created minimal test executable for debugging audio/visual synchronization and variable tempo system without full demo complexity. **Core Features**: Simple drum beat (kick-snare) with crash landmarks at bars 3 and 7, NOTE_A4 (440 Hz) reference tone at start of each bar for testing, screen flash effect synchronized to audio peaks, 16 second duration (8 bars at 120 BPM). **Variable Tempo Mode**: `--tempo` flag enables alternating tempo scaling (even bars: 1.0x → 1.5x acceleration, odd bars: 1.0x → 0.66x deceleration) to test music time vs physical time system. **Peak Logging**: `--log-peaks FILE` exports audio peak data for gnuplot visualization, `--log-peaks-fine` adds millisecond-resolution per-frame logging (~960 samples vs 32 beat-aligned samples). **Command-Line Options**: `--help` shows usage, `--fullscreen` for fullscreen mode, `--resolution WxH` for custom window size. **Error Handling**: Unknown options print error message and help text before exiting with status 1. **File Structure**: `src/test_demo.cc` (main executable, ~220 lines), `assets/test_demo.track` (drum patterns with NOTE_A4), `assets/test_demo.seq` (visual timeline), `test_demo_README.md` (comprehensive documentation). **Peak Log Format**: Beat-aligned mode logs at beat boundaries (32 samples), fine-grained mode logs every frame with beat_number column for correlation. **Build Integration**: CMake target with timeline/music generation, proper dependencies, size optimizations. **Use Cases**: Verify millisecond-precision sync, detect timing jitter, analyze tempo scaling effects, debug flash-to-audio alignment. **Impact**: Provides isolated testing environment for sync verification without demo64k complexity, enables data-driven analysis via exported logs.
-
-- **CMake Configuration Summary**: Added formatted configuration summary to CMake output displaying all build options (DEMO_SIZE_OPT, DEMO_STRIP_ALL, DEMO_BUILD_TESTS, etc.) with ON/OFF status, build type, and compiler information. Appears automatically at end of `cmake` command for easy verification of enabled options.
-
-- **Code Quality Fixes**: Replaced deprecated `std::sprintf` with `std::snprintf` in `asset_packer.cc` to eliminate compiler warnings and add buffer overflow protection during OBJ mesh vertex key generation.
-
-#### Milestone: Audio Playback Stability & Debug Infrastructure (February 4, 2026)
-- **Core Audio Backend Optimization**: Resolved critical audio playback issues (stop-and-go, glitches, eventual silence) caused by timing mismatches in miniaudio's Core Audio backend. Root cause: Core Audio optimized for 44.1kHz with 10ms periods, but our 32kHz system expected uniform ~13.78ms callbacks, causing resampling jitter. Fix: Added `allowNominalSampleRateChange = TRUE` to force OS-level 32kHz native and `performanceProfile = conservative` for larger buffers (4096 frames = 128ms). Result: Stable ~128ms callbacks with <1ms jitter, zero underruns during variable tempo (1.0x → 2.0x).
-
-- **Ring Buffer Capacity Tuning**: Increased ring buffer from 200ms to 400ms (25,600 samples) to handle tempo scaling headroom. Added comprehensive bounds checking with abort() on violations to catch buffer corruption early. Fixed critical bug: tempo-scaled rendering wasn't scaling dt when pre-filling buffer (`audio_render_ahead(g_music_time, dt * g_tempo_scale)`). Buffer now maintains 400ms fullness throughout playback, including 2.0x tempo acceleration.
-
-- **NOTE_ Parsing Bug Fix & Sample Caching**: Fixed critical `is_note_name()` bug that only checked first letter (A-G), causing ASSET_KICK_1 to be misidentified as A0 (27.5 Hz). Solution: Required "NOTE_" prefix (NOTE_E2, NOTE_G4) to distinguish from ASSET_* samples. Discovered resource exhaustion: every note event created NEW spectrogram (14 unique samples → 228 registrations). Implemented comprehensive caching in `tracker_init()`: pre-register all asset samples (loaded once from AssetManager) and generated notes (created once, stored in persistent pool). Result: 256 → 32 MAX_SPECTROGRAMS (88% memory reduction), zero spectrogram_id=-1 errors, perfect audio playback.
-
-- **Debug Logging Infrastructure**: Created systematic debug logging system (`src/util/debug.h`) with 7 category macros (DEBUG_LOG_AUDIO, DEBUG_LOG_RING_BUFFER, DEBUG_LOG_TRACKER, DEBUG_LOG_SYNTH, DEBUG_LOG_3D, DEBUG_LOG_ASSETS, DEBUG_LOG_GPU). Added `DEMO_ENABLE_DEBUG_LOGS` CMake option that defines `DEBUG_LOG_ALL` to enable all categories. Converted all diagnostic code in audio subsystem (miniaudio callbacks, ring buffer tracking, tracker validation, synth parameter checks) to use category macros. Default build: macros compile to `((void)0)` for zero runtime cost. Debug build: comprehensive logging (timing, buffer state, caching, validation). Updated `CONTRIBUTING.md` with pre-commit policy requiring debug build verification to ensure diagnostic code remains maintainable.
-
-- **Resource Analysis Tool**: Enhanced `tracker_compiler` to report required vs recommended pool sizes, cache potential, and memory usage. Analysis showed 152 required / 228 recommended spectrograms without caching, only 14 unique samples with caching. Tool now generates actionable optimization recommendations during music data compilation.
-
-#### Milestone: Audio System Robustness & Variable Tempo (February 4, 2026)
-- **Event-Based Tracker for Tempo Scaling**: Refactored tracker system from pattern compositing to individual event triggering. Previously, pattern events were baked into single spectrograms at fixed positions, so tempo scaling only affected pattern trigger timing. Now each TrackerEvent triggers as a separate voice with timing calculated dynamically based on music_time. Result: Notes within patterns correctly accelerate/decelerate with tempo changes. At 2.0x tempo, both pattern triggering AND note spacing play 2x faster. Verified with WAV dump showing 61.24s music time in 60s physical time during tempo transitions. All 17 tests pass (100%).
-
-- **WAV Dump Backend for Debugging**: Added `WavDumpBackend` implementing `AudioBackend` interface to render audio offline to .wav files instead of playing on device. Enabled via `--dump_wav output.wav` flag. Critical bug fix: Synth outputs STEREO (interleaved L/R) but initial implementation wrote MONO, causing severe distortion. Fixed by allocating `frames * 2` samples and writing stereo format (num_channels = 2). Added regression test (`test_wav_dump.cc`) with critical assertion `assert(header.num_channels == 2)` to prevent future mono/stereo mismatches. WAV format now matches live audio exactly: 16-bit PCM, stereo, 32kHz.
-
-- **Variable Tempo System**: Implemented unified music time abstraction in `main.cc` that decouples tracker timing from physical time. Music time advances at configurable `tempo_scale` rate (default 1.0), enabling dynamic tempo changes without pitch shifting. Created comprehensive test suite (`test_variable_tempo.cc`) verifying 2x speed-up and 2x slow-down "reset tricks" work perfectly. All 6 test scenarios pass with mathematical precision. System ready for expressive tempo control in demo with zero size impact.
-
-- **Task #51: Tracker Timing Verification System**: Created robust audio testing infrastructure with mock backend abstraction. Implemented `AudioBackend` interface separating synthesis from output, `MiniaudioBackend` for production, and `MockAudioBackend` for testing. Added event recording with precise timestamp tracking (32kHz sample rate). Created comprehensive test suite (`test_tracker_timing.cc`) that **verified simultaneous pattern triggers have 0.000ms delta** (perfect synchronization). All test infrastructure under `#if !defined(STRIP_ALL)` for zero production impact. 17/17 tests passing.
-
-#### Milestone: Audio Lifecycle Refactor (February 5, 2026)
-- **Task #56: AudioEngine Implementation**: Completed 4-phase refactor to eliminate fragile initialization order dependency between synth and tracker. **Phase 1 (Design & Prototype)**: Created `AudioEngine` class and `SpectrogramResourceManager` for unified lifecycle management with lazy loading. **Phase 2 (Test Migration)**: Migrated all tracker tests to use AudioEngine (test_tracker.cc, test_tracker_timing.cc, test_variable_tempo.cc, test_wav_dump.cc). **Phase 3 (Production Integration)**: Updated main.cc to use AudioEngine, fixed pre-existing procedural texture crash (flash_cube_effect.cc, hybrid_3d_effect.cc). **Phase 4 (Cleanup & Documentation)**: Removed backwards compatibility (synth_init from audio_init), updated HOWTO.md and CONTRIBUTING.md with usage patterns. Result: All 20 tests pass, binary size impact <500 bytes, initialization fragility eliminated.
-
-#### Milestone: Build System Optimization (February 6, 2026)
-- **Task C: CMake Dependency Graph Optimization**: Resolved critical build correctness bugs and improved developer iteration speed. **Header Split**: Refactored monolithic `asset_manager.h` (61 lines) into three focused headers: `asset_manager_dcl.h` (forward declarations for AssetId), `asset_manager.h` (core GetAsset/DropAsset API), and `asset_manager_utils.h` (typed helpers for TextureAsset/MeshAsset). Updated 17 source files to use appropriate headers. **Asset Dependency Tracking**: Implemented file-level dependency tracking for all 42 demo assets and 17 test assets. CMake now correctly tracks individual `.wgsl` shaders, `.spec` audio files, and `.obj` mesh files. **Critical Bug Fixed**: Changing shader files was NOT triggering asset regeneration, resulting in stale code in binaries. Developers had to manually `touch demo_assets.txt` as workaround. Now works correctly. **Performance**: Editing TextureAsset/MeshAsset helpers improved from 4.82s to 2.01s (58% faster). Shader edits now trigger correct 3.5s rebuild (was 0.28s with no rebuild - incorrect). **Implementation**: Added `parse_asset_list()` CMake function that parses `demo_assets.txt` format (`ASSET_NAME, COMPRESSION, FILENAME, DESC`) and extracts individual file paths for dependency tracking. All 20 tests pass. Zero functionality regressions.
-
-#### Milestone: Critical Shader Stability & Test Infrastructure (February 6, 2026) ✅
-- **Shader Crash Resolution**: Fixed three critical WGSL validation errors causing demo64k and test_3d_render to crash on startup. **Bug 1 (renderer_3d.wgsl)**: Removed dead code using non-existent `inverse()` function - WGSL doesn't provide matrix inverse, and validator checks all code paths even unreachable ones. Also removed reference to undefined `in.normal` vertex input. **Bug 2 (sdf_utils.wgsl, lighting.wgsl)**: Fixed `get_normal_basic()` function signature mismatch - changed from `obj_type: f32` to `obj_params: vec4<f32>` to match `get_dist()` calls. **Bug 3 (scene_query_linear.wgsl - ROOT CAUSE)**: Fixed linear scene query mode incorrectly declaring binding 2 (BVH storage buffer). Linear version was identical to BVH version due to copy-paste error. Replaced BVH traversal with proper linear object iteration loop. **Impact**: When `use_bvh=false`, pipeline created without binding 2, but shader expected it → validation error → crash.
-
-- **Comprehensive Shader Compilation Tests**: Created `test_shader_compilation.cc` to prevent regression. Test compiles all production shaders through WebGPU (`wgpuDeviceCreateShaderModule`), validates both BVH and Linear composition modes using `ShaderComposer`, catches WGSL syntax errors, binding mismatches, type errors, and function signature issues. **Test Gap Analysis**: Existing `test_shader_assets.cc` only checked for keywords (`@vertex`, `fn`, etc.) without actual compilation - would NOT have caught any of the three bugs. New test fills this gap with real GPU validation. **Graceful Degradation**: Test handles platforms where WebGPU device unavailable by skipping GPU tests but still validating shader composition.
-
-- **Results**: demo64k runs cleanly without WebGPU errors, test_3d_render no longer crashes, 22/23 tests pass (FftTest unrelated to this work), comprehensive regression prevention infrastructure in place. All shader composition modes (BVH/Linear) now validated in CI. Files modified: renderer_3d.wgsl, sdf_utils.wgsl, lighting.wgsl, scene_query_linear.wgsl, test_shader_compilation.cc (new), CMakeLists.txt.
-
-#### Milestone: FFT-based DCT/IDCT Complete (February 6, 2026) ✅
-- **Core FFT Implementation**: Replaced failing double-and-mirror method with Numerical Recipes reordering method for DCT via FFT. Fixed reference IDCT to use DCT-III (inverse of DCT-II, not IDCT-II). All transforms now use orthonormal normalization: sqrt(1/N) for DC term, sqrt(2/N) for AC terms. Result: Perfect round-trip accuracy for impulse signals, <5e-3 error for sinusoidals (acceptable for FFT).
-
-- **Audio Pipeline Integration**: Integrated FFT-based DCT/IDCT into audio engine (`src/audio/idct.cc`, `fdct.cc`) and both web editors (`tools/spectral_editor/dct.js`, `tools/editor/dct.js`). All synthesis paths now use fast O(N log N) FFT instead of O(N²) naive DCT. Regenerated all 14 spectrogram assets with orthonormal DCT to match new synthesis engine.
-
-- **Critical Fixes**: **(1) Normalization Mismatch**: Old non-orthonormal DCT produced 16× larger values. Solution: Regenerated all `.spec` assets with orthonormal DCT. **(2) Procedural Notes**: NOTE_* generation inaudible after normalization change. Solution: Added sqrt(DCT_SIZE/2) = 16× scaling compensation in `gen.cc`. **(3) Windowing Error**: Hamming window incorrectly applied to spectrum before IDCT (should only be used for analysis, not synthesis). Solution: Removed windowing from `synth.cc` and both editors. Result: Correct volume, no distortion, clean frequency spectrum.
-
-- **Testing & Verification**: All 23 tests pass (100% success rate). Round-trip accuracy verified (impulse at index 0: perfect). Audio playback: correct volume, no distortion. Procedural notes: audible at correct levels. Web editors: clean spectrum, no comb artifacts. WAV dumps match expected output.
-
-- **Key Technical Insights**: (1) DCT-III is the inverse of DCT-II, not IDCT-II. (2) Hamming window is ONLY for analysis (before DCT), NOT synthesis (before IDCT). (3) Orthonormal DCT produces sqrt(N/2) smaller values than non-orthonormal. (4) Reordering method is more accurate than double-and-mirror for DCT via FFT. (5) Round-trip accuracy is more important than absolute DCT accuracy.
-
-#### Milestone: Interactive Timeline Editor (February 5, 2026) 🎉
-- **Task #57 Phase 1: Production-Ready Timeline Editor**: Created fully functional web-based editor for `demo.seq` timeline files. **Core Features**: Load/save demo.seq with BPM parsing, Gantt-style visual timeline, drag & drop sequences/effects with snap-to-beat, resize handles (left/right) allowing negative relative times, stack-order based priority system (Up/Down buttons + "Same as Above" toggle), floating auto-apply properties panel, diagonal mouse wheel scroll with 10% viewport slack, dynamic sequence bounds calculation, delete/add sequences, re-order by time. **Audio Visualization**: WAV waveform display above timeline using Web Audio API, scales with zoom (pixels per second), documented integration with `--dump_wav` flag for aligning sequences with actual demo audio output. **UI Polish**: Hoverable sequence names (large centered, fades on hover), hidden scrollbars, crosshair cursor on waveform, flash animation on active sequence change, clean minimal interface. **Bug Fixes**: Resolved critical e.target vs e.currentTarget drag offset bug, fixed sequence overlap with cumulative Y positioning, corrected effect relative time calculations. **Files**: `tools/timeline_editor/index.html` (~1200 lines pure HTML/CSS/JS, no dependencies), `README.md` (usage guide, wav_dump integration docs), `ROADMAP.md` (3 phases, 117-161 hour estimate). **Next**: Phase 1.2 (Add Effect button), Phase 2.5 (music.track visualization overlay). **Impact**: Visual timeline editing now production-ready, eliminates manual text editing for sequence placement and timing adjustments.
-
-- **Task #50: WGSL Modularization**: Updated `ShaderComposer` to support recursive `#include` directives, refactored the entire shader library into granular snippets (shapes, utils, lighting), and updated the 3D renderer to use this modular system. This resolved macOS shader compilation issues and significantly improved shader maintainability.
-- **Task #48: Improve Audio Coverage**: Achieved 93% coverage for `src/audio/` by adding dedicated tests for DCT transforms, procedural generation, and synthesis rendering.
-- **Task #47: Improve Asset Manager Coverage**: Increased `asset_manager.cc` coverage to 88% by testing runtime error paths (unknown functions, generation failure).
-- **Task #46: Enhance Coverage Script**: Updated coverage report script to support directory filtering (e.g., `./scripts/gen_coverage_report.sh src/procedural`).
-- **Task #45: Improve Procedural Generation Coverage**: Achieved 96% coverage for `src/procedural/` by implementing comprehensive tests for Perlin noise, periodic blending, and parameter handling.
-- **Task #44: Developer Tooling (Coverage)**: Added `DEMO_ENABLE_COVERAGE` CMake option and created `scripts/gen_coverage_report.sh` to generate HTML coverage reports using `lcov` on macOS.
-- **Skybox & Two-pass Rendering Stability**: Resolved "black screen" and validation errors by implementing a robust two-pass rendering architecture (Pass 1: Skybox/Clear, Pass 2: Scene Objects). Implemented a rotating skybox using world-space ray unprojection (`inv_view_proj`) and a multi-octave procedural noise generator.
-- **Task #20: Platform & Code Hygiene**: Consolidated platform-specific shims and WebGPU headers into `platform.h`. Refactored `platform_init` and `platform_poll` for better abstraction. Removed STL containers from initial hot paths (`AssetManager`, `procedural`). Full STL removal for CRT replacement is deferred to the final optimization phase.
-- **Task #26: Shader Asset Testing & Validation**: Developed comprehensive tests for `ShaderComposer` and WGSL asset loading/composition. Added a shader validation test to ensure production assets are valid.
-- **Asset Pipeline Improvement**: Created a robust `gen_spectrograms.sh` script to automate the conversion of `.wav` and `.aif` files to `.spec` format, replacing the old, fragile script. Added 13 new drum and bass samples to the project.
-- **Build System Consolidation (Task #25)**: Modularized the build by creating subsystem libraries (audio, gpu, 3d, util, procedural) and implemented helper macros to reduce boilerplate in `CMakeLists.txt`. This improves build maintenance and prepares for future CRT replacement.
-- **Asset System Robustness**: Resolved "static initialization order fiasco" by wrapping the asset table in a "Construct On First Use" getter (`GetAssetRecordTable()`), ensuring assets are available during dynamic global initialization (e.g., shader strings).
-- **Shader Asset Integration (Task #24)**: Extracted all hardcoded WGSL strings into `.wgsl` assets, registered them in `demo_assets.txt`, and updated `Renderer3D`, `VisualDebug`, and `Effects` to use `GetAsset` and `ShaderComposer`.
-- **WebGPU Stabilization**: Resolved `WGPUSurface` creation failures on macOS by adding platform-specific `GLFW_EXPOSE_NATIVE_COCOA` definitions and fixed validation errors in the render pass configuration.
-- **Final Build Stripping (Task #8)**: Implemented the `STRIP_ALL` macro to remove non-essential code (CLI parsing, debug labels, iostream) and refined size optimization flags (`-dead_strip`) for macOS.
-- **Minimal Audio Tracker (Task 21.3)**: Finalized a pattern-based audio tracker supporting both procedural notes and asset-based spectrograms with a unified "one-voice-per-pattern" pasting strategy.
-- **WGSL Library (Task 21.1)**: Implemented `ShaderComposer` for modular WGSL snippet management.
-- **Tight Ray Bounds (Task 21.2)**: Implemented local-space ray-box intersection to optimize SDF raymarching.
-- **High-DPI Fix**: Resolved viewport "squishing" via dynamic resolution uniforms and explicit viewports.
-- **Unified 3D Shadows**: Implemented robust SDF shadows across all objects using `inv_model` transforms.
-- **test_mesh tool**: Implemented a standalone `test_mesh` tool for visualizing OBJ files with debug normal display.
+### Current Status
+- Audio system: Stable with real-time peak tracking, variable tempo support, comprehensive test coverage
+- Build system: Optimized with proper asset dependency tracking
+- Shader system: Modular with comprehensive compilation tests
+- 3D rendering: Hybrid SDF/rasterization with BVH acceleration
 
 ---
 ## Next Up
diff --git a/TODO.md b/TODO.md
index 83f6d35..000e619 100644
--- a/TODO.md
+++ b/TODO.md
@@ -2,178 +2,7 @@
 
 This file tracks prioritized tasks with detailed attack plans.
 
-## Recently Completed (February 7, 2026)
-
-- [x] **Audio Peak Measurement & Test Coverage Improvements** (February 7, 2026)
-    - [x] **Real-Time Peak Fix**: Fixed critical audio-visual sync bug where visual effects triggered ~400ms before audio was heard
-      - Root cause: Peak measured at ring buffer write time (synth_render) instead of playback time (audio callback)
-      - Solution: Added `get_realtime_peak()` to AudioBackend interface, implemented in MiniaudioBackend audio_callback
-      - Used exponential averaging: instant attack, 0.7 decay rate (1-second fade time)
-    - [x] **Peak Decay Optimization**: Fixed "test_demo just flashing" issue
-      - Old: 0.95 decay = 5.76 second fade (screen stayed white)
-      - New: 0.7 decay = 1.15 second fade (proper flash with smooth decay)
-    - [x] **SilentBackend Creation**: Test-only backend for audio.cc testing without hardware
-      - Created src/audio/backend/silent_backend.{h,cc}
-      - 7 comprehensive tests: lifecycle, peak control, tracking, playback time, buffer management
-      - Significantly improved audio.cc test coverage
-    - [x] **Backend Reorganization**: Moved all backends to src/audio/backend/ subdirectory
-      - Cleaner code organization, updated all includes and CMake paths
-    - [x] **Dead Code Removal**: Removed unused `register_spec_asset()` function
-    - **Result**: All 28 tests passing, perfect audio-visual sync, improved test coverage
-
-## Recently Completed (February 6, 2026)
-
-- [x] **Critical Shader Bug Fixes & Test Infrastructure** (February 6, 2026)
-    - [x] **Shader Validation Errors**: Fixed three critical WGSL bugs causing demo64k and test_3d_render crashes:
-      - Removed dead code using non-existent `inverse()` function in renderer_3d.wgsl
-      - Fixed `get_normal_basic()` signature mismatch in sdf_utils.wgsl and lighting.wgsl (obj_type: f32 → obj_params: vec4<f32>)
-      - Fixed scene_query_linear.wgsl incorrectly declaring binding 2 (BVH buffer) - was identical to BVH version due to copy-paste error
-    - [x] **Root Cause**: Linear shader mode expected no binding 2, but shader declared it, causing pipeline/shader mismatch
-    - [x] **New Test Coverage**: Created `test_shader_compilation.cc` that compiles all production shaders through WebGPU, tests both BVH and Linear composition modes, validates WGSL syntax/bindings/types
-    - [x] **Test Gap Analysis**: Existing test_shader_assets only checked keywords, not actual compilation. New test would have caught all three bugs.
-    - **Result**: demo64k runs without WebGPU errors, test_3d_render no longer crashes, 22/23 tests pass (FftTest unrelated), comprehensive regression prevention
-
-- [x] **Task C: Build System Optimization** (February 6, 2026)
-    - [x] **Header Split**: Refactored `asset_manager.h` into `asset_manager_dcl.h` (forward declarations), `asset_manager.h` (core API), and `asset_manager_utils.h` (typed helpers for TextureAsset/MeshAsset).
-    - [x] **Asset Dependency Tracking**: Added file-level dependencies for all assets (.wgsl shaders, .spec audio, .obj meshes). CMake now tracks 42 demo assets + 17 test assets individually.
-    - [x] **Performance Impact**: Editing TextureAsset/MeshAsset helpers: 4.82s → 2.01s (58% faster). Editing shaders now triggers correct rebuild (was broken before - stale code bug fixed).
-    - [x] **Developer Workflow**: No more manual `touch demo_assets.txt` workaround needed. Edit any asset file and rebuild works correctly.
-    - **Result**: Critical correctness bug fixed (shader changes were not triggering rebuilds). Build system now has proper dependency tracking for all source and asset files.
-
-## Recently Completed (February 5, 2026)
-
-- [x] **Audio Lifecycle Refactor (Task #56)**:
-    - [x] **Phase 1: Design & Prototype**: Created `AudioEngine` class and `SpectrogramResourceManager` to manage audio subsystem initialization and resource loading with lazy loading strategy.
-    - [x] **Phase 2: Test Migration**: Migrated all tracker-related tests (`test_tracker.cc`, `test_tracker_timing.cc`, `test_variable_tempo.cc`, `test_wav_dump.cc`) to use AudioEngine instead of direct synth/tracker initialization.
-    - [x] **Phase 3: Production Integration**: Updated `main.cc` to use AudioEngine, eliminating initialization order fragility in production code. Fixed pre-existing demo crash (procedural texture loading).
-    - [x] **Phase 4: Cleanup & Documentation**: Removed backwards compatibility (synth_init() from audio_init()), updated HOWTO.md and CONTRIBUTING.md with AudioEngine usage patterns and best practices.
-    - **Result**: Initialization order dependency eliminated. All 20 tests pass. Binary size impact <500 bytes. Demo runs successfully.
-
-- [x] **Physics & Collision (Task #49)**:
-    - [x] **CPU-Side SDF Library**: Implemented `sdSphere`, `sdBox`, `sdTorus`, `sdPlane` in `src/3d/sdf_cpu.h` using `mini_math.h`. Added `calc_normal` for numerical gradients.
-    - [x] **BVH Construction**: Implemented `BVHNode` and `BVHBuilder` in `src/3d/bvh.h/cc`. Optimized broad-phase collision queries with `BVH::Query`.
-    - [x] **Physics System**: Created `PhysicsSystem` in `src/3d/physics.h/cc` implementing semi-implicit Euler integration, broad-phase BVH culling, and narrow-phase SDF proxy probing.
-    - [x] **Integration & Visualization**: Added `velocity`, `mass`, `is_static` to `Object3D`. Integrated physics loop into `test_3d_render.cc` and added BVH wireframe visualization to `VisualDebug`.
-
-- [x] **Audio Playback Debugging & Core Audio Optimization**:
-    - [x] **Core Audio Timing Fix**: Resolved stop-and-go audio glitches caused by timing mismatch. Core Audio optimized for 44.1kHz (10ms periods), but our 32kHz system expected uniform ~13.78ms callbacks, causing resampling jitter. Fix: Added `allowNominalSampleRateChange = TRUE` to force OS-level 32kHz native and `performanceProfile = conservative` for 4096-frame buffers (128ms). Result: Stable ~128ms callbacks, <1ms jitter, zero underruns.
-    - [x] **Ring Buffer Capacity**: Increased from 200ms to 400ms (25,600 samples) to handle tempo scaling headroom. Added comprehensive bounds checking with abort() on violations.
-    - [x] **Tempo-Scaled Buffer Fill**: Fixed critical bug where buffer pre-fill didn't scale dt by tempo (`audio_render_ahead(g_music_time, dt * g_tempo_scale)`). Buffer now maintains 400ms fullness during 2.0x acceleration.
-    - [x] **Extensive Diagnostics**: Added high-resolution timing tracking (clock_gettime), callback interval measurement, buffer level monitoring, underrun detection. All under conditional compilation for future debugging.
-
-- [x] **NOTE_ Parsing Fix & Sample Caching**:
-    - [x] **Parser Bug Fix**: Fixed `is_note_name()` checking only first letter (A-G), causing ASSET_KICK_1 → A0 (27.5 Hz) false positives. Required "NOTE_" prefix to distinguish notes from assets. Updated music.track to use NOTE_E2, NOTE_G4 format.
-    - [x] **Resource Exhaustion Discovery**: Found every event created NEW spectrogram (14 unique samples → 228 registrations). MAX_SPECTROGRAMS=16 insufficient, causing spectrogram_id=-1 errors.
-    - [x] **Comprehensive Caching**: Implemented cache in `tracker_init()` pre-registering all samples. Assets: loaded once from AssetManager. Generated notes: created once, stored in persistent pool. All cached synth_ids reused.
-    - [x] **Memory Reduction**: MAX_SPECTROGRAMS reduced from 256 (temporary workaround) to 32 (2.3x headroom over 14 actual). 88% memory savings vs uncached approach.
-    - [x] **Resource Analysis Tool**: Enhanced `tracker_compiler` to report required/recommended pool sizes, cache potential, memory usage. Analysis showed 152/228 required without caching, 14 with caching.
-
-- [x] **Debug Logging Infrastructure**:
-    - [x] **Central Debug Header**: Created `src/util/debug.h` with 7 category macros (DEBUG_LOG_AUDIO, DEBUG_LOG_RING_BUFFER, DEBUG_LOG_TRACKER, DEBUG_LOG_SYNTH, DEBUG_LOG_3D, DEBUG_LOG_ASSETS, DEBUG_LOG_GPU).
-    - [x] **CMake Integration**: Added `DEMO_ENABLE_DEBUG_LOGS` option defining `DEBUG_LOG_ALL` to enable all categories. Individual categories can be enabled selectively.
-    - [x] **Source Code Conversion**: Updated miniaudio_backend.cc (timing, device config), ring_buffer.cc (underruns), tracker.cc (validation), synth.cc (parameter checks) to use category macros.
-    - [x] **Zero Runtime Cost**: Default build: macros compile to `((void)0)`. Debug build: comprehensive logging preserved for troubleshooting.
-    - [x] **Pre-Commit Policy**: Updated CONTRIBUTING.md requiring debug build verification before significant commits to ensure diagnostic code remains maintainable.
-    - [x] **Verification**: Both default and debug builds compile without errors. Audio playback works correctly in both modes.
-
-- [x] **Event-Based Tracker for Tempo Scaling**:
-    - [x] **Problem Identified**: Notes within patterns didn't accelerate with tempo changes. Pattern events were pre-composited into single spectrograms at fixed positions.
-    - [x] **Refactored Architecture**: Changed from pattern compositing to individual event triggering. Each TrackerEvent now triggers as separate voice.
-    - [x] **Dynamic Beat Calculation**: `elapsed_beats = (music_time - start_time) / beat_duration` allows notes to respect tempo scaling.
-    - [x] **ActivePattern Tracking**: Added structure to track pattern_id, start_music_time, and next_event_idx for each active pattern instance.
-    - [x] **Removed Compositing Logic**: Deleted paste_spectrogram approach that baked events at fixed frame offsets.
-    - [x] **Full Tempo Scaling**: At 2.0x tempo, both pattern triggering AND note spacing play 2x faster. At 0.5x tempo, both play 2x slower.
-    - [x] **Testing**: Updated test_tracker.cc to verify individual event triggers at specific beat times. All 17 tests pass.
-    - [x] **WAV Verification**: Confirmed with WAV dump showing 61.24s music time in 60s physical time during tempo transitions.
-
-- [x] **WAV Dump Backend for Debugging**:
-    - [x] **Implementation**: Created WavDumpBackend implementing AudioBackend interface to render audio offline to .wav files.
-    - [x] **Command-Line Option**: Added `--dump_wav output.wav` flag to enable offline rendering instead of live playback.
-    - [x] **Stereo Bug Fix**: Fixed critical mono/stereo mismatch. Synth outputs STEREO (interleaved L/R) but initial implementation wrote MONO.
-    - [x] **Correct Allocation**: Changed to allocate `frames * 2` samples and write stereo format (num_channels = 2).
-    - [x] **Format Matching**: WAV header now correctly specifies 16-bit PCM, stereo, 32kHz (matches live audio exactly).
-    - [x] **Regression Test**: Added test_wav_dump.cc with critical assertion `assert(header.num_channels == 2)` to prevent future mismatches.
-    - [x] **Tempo Simulation**: WAV dump backend simulates tempo scaling matching main.cc logic for accurate offline rendering.
-    - [x] **All Tests Pass**: 17/17 tests pass including WAV format verification.
-
-- [x] **Variable Tempo System**:
-    - [x] **Music Time Abstraction**: Implemented unified music time in `main.cc` that advances at `tempo_scale` rate, decoupling from physical time.
-    - [x] **Tempo Control**: Added `g_tempo_scale` (default 1.0) allowing future dynamic tempo changes without pitch shifting.
-    - [x] **Reset Tricks**: Comprehensive tests verify 2x speed-up and 2x slow-down reset techniques work correctly.
-    - [x] **Test Suite**: Created `test_variable_tempo.cc` with 6 test scenarios: basic scaling, speed-up/slow-down resets, pattern density swap, continuous acceleration, oscillating tempo.
-    - [x] **Perfect Verification**: All tests pass, confirming music_time advances correctly at variable rates (e.g., 2.0x tempo → 2x faster triggering).
-    - [x] **Zero Pitch Shift**: Spectrograms unchanged, only trigger timing affected (as designed).
-    - [x] **Documentation**: Created `ANALYSIS_VARIABLE_TEMPO_V2.md` explaining simplified trigger-timing approach.
-
-- [x] **Task #51.3 & #51.4: Tracker Test Suite & Build Integration**:
-    - [x] **Comprehensive Test Suite**: Created `test_tracker_timing.cc` with 7 test scenarios using MockAudioBackend.
-    - [x] **Simultaneous Trigger Verification**: Confirmed multiple patterns at same time trigger with **0.000ms delta** (perfect sync).
-    - [x] **Test Coverage**: Basic recording, progressive triggering, simultaneous triggers, monotonicity, seek/fast-forward, timestamp clustering, render integration.
-    - [x] **Real Music Data**: Tests use generated tracker music data for realistic validation.
-    - [x] **Build Integration**: Added to CMake with proper dependencies on generated music data.
-    - [x] **All Tests Pass**: 15/15 tests passing (100% success rate).
-
-- [x] **Task #51.2: Mock Audio Backend**:
-    - [x] **Event Recording**: Created `VoiceTriggerEvent` structure to capture timestamp, spectrogram_id, volume, and pan.
-    - [x] **MockAudioBackend Class**: Implemented test-only backend with event recording and time tracking capabilities.
-    - [x] **Time Management**: Added `advance_time()`, `set_time()`, and `get_current_time()` for deterministic testing.
-    - [x] **Frame Rendering Hook**: Implemented `on_frames_rendered()` to automatically update time based on audio frames (32kHz).
-    - [x] **Synth Integration**: Verified mock backend correctly captures voice triggers from synth engine.
-    - [x] **Comprehensive Tests**: Created `test_mock_backend.cc` with 6 test scenarios covering all mock functionality.
-    - [x] **Build Integration**: Added mock backend to test builds, all 14 tests pass.
-
-- [x] **Task #51.1: Audio Backend Abstraction**:
-    - [x] **Interface Created**: Defined `AudioBackend` interface in `src/audio/audio_backend.h` with hooks for voice triggering and frame rendering.
-    - [x] **Production Backend**: Moved miniaudio implementation from `audio.cc` to `MiniaudioBackend` class, maintaining backward compatibility.
-    - [x] **Audio Refactoring**: Updated `audio.cc` to use backend abstraction with automatic fallback to `MiniaudioBackend`.
-    - [x] **Event Hooks**: Added time tracking to `synth.cc` with `on_voice_triggered()` callbacks (guarded by `!STRIP_ALL`).
-    - [x] **Verification Test**: Created `test_audio_backend.cc` to verify backend injection and event recording work correctly.
-    - [x] **Build Integration**: Updated CMakeLists.txt to include new backend files and link audio tests properly with util/procedural dependencies.
-    - [x] **Zero Size Impact**: All test infrastructure under `#if !defined(STRIP_ALL)`, production path unchanged.
-
-- [x] **Task #50: WGSL Modularization**:
-    - [x] **Recursive Composition**: Updated `ShaderComposer` to support recursive `#include "snippet_name"` directives with cycle detection.
-    - [x] **Granular SDF Library**: Extracted `math/sdf_shapes.wgsl`, `math/sdf_utils.wgsl`, `render/shadows.wgsl`, `render/scene_query.wgsl`, and `render/lighting_utils.wgsl`.
-    - [x] **Pipeline Update**: Refactored `Renderer3D` and `renderer_3d.wgsl` to use the new modular system, reducing C++-side dependency management.
-    - [x] **Platform Fix**: Resolved `WGPUShaderSourceWGSL` usage on macOS to ensure compatibility with composed shader strings.
-
-- [x] **Task #48: Improve Audio Coverage**:
-    - [x] **New Tests**: Added `test_dct` (100% coverage for transforms) and `test_audio_gen` (94% coverage for procedural audio).
-    - [x] **Enhanced Tests**: Updated `test_synth` to cover rendering loop, double-buffering, and resource exhaustion.
-    - [x] **Coverage Boost**: Increased `src/audio/` coverage from ~42% to 93%.
-- [x] **Task #47: Improve Asset Manager Coverage**:
-    - [x] **New Tests**: Added tests for unknown procedural functions, generation failures, and edge cases in `src/tests/test_assets.cc`.
-    - [x] **Tooling Update**: Downgraded `asset_packer` validation error to warning to allow testing invalid assets.
-    - [x] **Coverage Boost**: Increased `src/util/asset_manager.cc` coverage from 71% to 88%.
-- [x] **Task #46: Enhance Coverage Script**: Updated `scripts/gen_coverage_report.sh` to accept an optional directory argument for targeted coverage reports (e.g., `src/procedural`).
-- [x] **Task #45: Improve Procedural Generation Coverage**:
-    - [x] **Unit Tests:** Implemented comprehensive tests for `gen_perlin`, `make_periodic`, and default parameter handling in `src/tests/test_procedural.cc`.
-    - [x] **Coverage Boost:** Increased `src/procedural/generator.cc` coverage from 38% to 96%.
-- [x] **Task #44: Developer Tooling (Coverage)**:
-    - [x] **Implement Code Coverage:** Added `DEMO_ENABLE_COVERAGE` CMake option and created `scripts/gen_coverage_report.sh` to generate HTML coverage reports using `lcov` on macOS.
-    - [x] **Documentation:** Updated `doc/HOWTO.md` with usage instructions.
-- [x] **Skybox & Two-pass Rendering Stability**:
-    - [x] **Fixed Two-pass Rendering:** Implemented mandatory clear operations for color and depth when the skybox is absent, preventing black screens and depth validation errors.
-    - [x] **Implemented Rotating Skybox:** Added `inv_view_proj` to `GlobalUniforms` and updated the skybox shader to perform world-space ray unprojection (`inv_view_proj`), enabling correct rotation with the camera.
-    - [x] **Enhanced Procedural Noise:** Implemented a multi-octave Value Noise generator for higher-quality skybox textures.
-    - [x] **Scene Integrity:** Restored proper object indexing and removed redundant geometry, ensuring the floor grid and objects render correctly.
-
-- [x] **Task #57: Interactive Timeline Editor (Phase 1 Complete)** 🎉
-    - [x] **Core Parser & Renderer**: Implemented demo.seq parser with BPM, beat notation, priority modifiers. Gantt-style timeline rendering with dynamic sequence/effect positioning.
-    - [x] **Drag & Drop**: Sequences and effects draggable along timeline with proper offset calculation. Fixed critical e.target vs e.currentTarget bug preventing erratic jumping.
-    - [x] **Resize Handles**: Left/right handles on selected effects. Allow negative relative times (effects extend before sequence start).
-    - [x] **Snap-to-Beat**: Checkbox toggle with beat markers. Automatic snapping when dragging in beat mode.
-    - [x] **Properties Panel**: Floating, collapsible panel with auto-apply (no Apply button). Properties update on input change.
-    - [x] **Stack-Order Priority**: Up/Down buttons to reorder effects within sequence. Stack position determines rendering priority (higher index = rendered later = on top).
-    - [x] **Priority Modifiers**: Toggle button for "Same as Above" (= modifier) vs "Increment" (+ modifier). Visual feedback in button text.
-    - [x] **Diagonal Scroll**: Mouse wheel navigation with 10% viewport slack. Smooth following to time-ordered sequence cascade. Flash animation on active sequence change.
-    - [x] **Dynamic Bounds**: Sequence visual bounds calculated from min/max effect times. Cumulative Y positioning prevents overlap.
-    - [x] **Waveform Visualization**: Load WAV files (via Web Audio API) and display waveform above timeline. Scales with zoom (pixels per second). Documented integration with `--dump_wav` flag.
-    - [x] **UI Polish**: Hoverable sequence names (large centered text, fades on hover). No scrollbars (hidden CSS). Clean, minimal interface.
-    - [x] **File I/O**: Load/save demo.seq files with proper serialization. BPM parsing and display. Re-order sequences by time.
-    - [x] **Delete & Add**: Delete sequences/effects. Add new sequences (at time 0, priority 0).
-    - **Result**: Fully functional timeline editor ready for production use. Phase 1.1 complete (basic editing, snap-to-beat, waveform). Phase 1.2 (Add Effect button) and Phase 2.5 (music.track visualization) documented in ROADMAP.md.
-    - **Files**: `tools/timeline_editor/index.html` (~1200 lines), `README.md`, `ROADMAP.md` (3 phases, 117-161 hour estimate).
+**Note:** For a history of recently completed tasks, see `COMPLETED.md`.
 
 ## Critical Fixes
 
@@ -287,11 +116,9 @@ This file tracks prioritized tasks with detailed attack plans.
 ## Priority 4: Developer Tooling & CI
 **Goal**: Improve developer workflows, code quality, and release processes.
 
-- [ ] **Task #51: Tracker Timing Verification**
-    - [x] **Task #51.1: Audio Backend Abstraction**: Create an interface to separate audio output from synth logic, enabling testable backends.
-    - [x] **Task #51.2: Mock Audio Backend**: Implement a test backend that records voice trigger events with precise timestamps.
-    - [x] **Task #51.3: Tracker Test Suite**: Create `test_tracker.cc` to verify pattern triggering, timing accuracy, and synchronization.
-    - [x] **Task #51.4: Integration with Build**: Wire up tests to CMake and ensure they run in CI.
+- [x] **Task #51: Tracker Timing Verification** ✅ COMPLETED
+    - Created robust audio testing infrastructure with mock backend abstraction
+    - All subtasks complete, moved to COMPLETED.md
 
 ## Phase 2: Size Optimization (Final Goal)
 
@@ -451,207 +278,4 @@ This file tracks prioritized tasks with detailed attack plans.
 
 ---
 
-## Task #51: Tracker Timing Verification - Detailed Attack Plan
-
-**Problem Statement**: The tracker and synthesizer have audio sync issues. There's no robust way to verify that tracker patterns trigger at the correct timestamps without running the full audio hardware stack.
-
-**Goal**: Implement a testable audio backend abstraction with event recording capabilities to verify tracker timing accuracy.
-
-### Task #51.1: Audio Backend Abstraction
-**Objective**: Decouple audio output from synthesis logic to enable testing without hardware.
-
-**Implementation Steps**:
-- [ ] **Create `src/audio/audio_backend.h`**:
-  - Define `AudioBackend` interface with pure virtual methods:
-    - `init()`: Initialize backend resources
-    - `start()`: Start audio playback/recording
-    - `shutdown()`: Clean up resources
-    - `on_voice_triggered(timestamp, spec_id, volume, pan)`: Hook for voice events
-  - Add `#if !defined(STRIP_ALL)` guards around test-only methods
-
-- [ ] **Create `src/audio/miniaudio_backend.h` and `.cc`**:
-  - Move current miniaudio implementation from `audio.cc` to `MiniaudioBackend` class
-  - Implement `AudioBackend` interface
-  - Keep production behavior identical (no regressions)
-  - This backend does NOT record events (production path)
-
-- [ ] **Refactor `src/audio/audio.cc`**:
-  - Add global `AudioBackend* g_audio_backend` pointer
-  - Add `void audio_set_backend(AudioBackend* backend)` function (under `!STRIP_ALL`)
-  - Default to `MiniaudioBackend` if no backend is set
-  - Replace direct miniaudio calls with backend interface calls
-
-- [ ] **Update `src/audio/synth.cc`**:
-  - Add external hook: `extern AudioBackend* g_audio_backend_for_events`
-  - In `synth_trigger_voice()`, call `backend->on_voice_triggered()` if backend exists
-  - Ensure hook is `#if !defined(STRIP_ALL)` guarded
-
-**Size Impact**: Zero (test code stripped in final build).
-
-**Validation**: Existing tests (`test_synth.cc`) must pass unchanged.
-
----
-
-### Task #51.2: Mock Audio Backend
-**Objective**: Create a test-only backend that records all audio events with timestamps.
-
-**Implementation Steps**:
-- [ ] **Create `src/audio/mock_audio_backend.h` and `.cc`** (under `#if !defined(STRIP_ALL)`):
-  - Define `struct VoiceTriggerEvent`:
-    ```cpp
-    struct VoiceTriggerEvent {
-        float timestamp_sec;
-        int spectrogram_id;
-        float volume;
-        float pan;
-    };
-    ```
-  - Implement `MockAudioBackend` class:
-    - Maintain `std::vector<VoiceTriggerEvent> recorded_events`
-    - Override `on_voice_triggered()` to record events with current time
-    - Add `const std::vector<VoiceTriggerEvent>& get_events() const`
-    - Add `void clear_events()`
-    - Add `void advance_time(float delta_sec)` to simulate time progression
-    - Implement `init()`, `start()`, `shutdown()` as no-ops
-
-- [ ] **Time Tracking**:
-  - Add `float current_time_sec` member to `MockAudioBackend`
-  - Increment `current_time_sec` in `advance_time()`
-  - Use `current_time_sec` as timestamp when recording events
-
-- [ ] **Synth Integration**:
-  - When `synth_render()` is called, calculate frames rendered
-  - Notify backend of time elapsed: `frames / sample_rate`
-  - Update mock's internal clock accordingly
-
-**Testing**: Create minimal unit test in `test_tracker.cc` to verify event recording works.
-
----
-
-### Task #51.3: Tracker Test Suite
-**Objective**: Comprehensive tests for tracker pattern triggering and timing accuracy.
-
-**Implementation Steps**:
-- [ ] **Create `src/tests/test_tracker.cc`**:
-
-  - **Test 1: Single Pattern Trigger**
-    - Define a minimal `TrackerScore` with 1 pattern at `t=1.0s`
-    - Set up mock backend
-    - Call `tracker_update(0.5)` → verify no events
-    - Call `tracker_update(1.0)` → verify 1 event recorded
-    - Validate event timestamp matches expected trigger time
-
-  - **Test 2: Multiple Pattern Triggers**
-    - Score with 3 patterns at `t=0.5s, 1.0s, 2.0s`
-    - Progressively call `tracker_update()` with increasing times
-    - Verify each pattern triggers exactly once at correct time
-
-  - **Test 3: Event Timing Accuracy**
-    - Pattern with multiple events at different beat offsets
-    - Verify each event's timestamp matches: `pattern_start_time + (beat * beat_duration)`
-    - Use tolerance: `±1 frame` (1/32000 sec ≈ 31.25µs)
-
-  - **Test 4: BPM Scaling**
-    - Same pattern tested at different BPMs (60, 120, 180)
-    - Verify beat-to-time conversion is accurate: `beat_sec = 60.0 / bpm`
-
-  - **Test 5: Pattern Overlap**
-    - Two patterns with overlapping time ranges
-    - Verify both trigger correctly without interference
-
-  - **Test 6: Asset vs Procedural Samples**
-    - Pattern using both asset-based spectrograms and procedural notes
-    - Verify both types render and trigger correctly
-
-  - **Test 7: Seek/Fast-Forward Simulation**
-    - Simulate `audio_render_silent()` behavior
-    - Start at `t=0`, fast-forward to `t=10.0s`
-    - Verify all patterns in range [0, 10] triggered correctly
-
-- [ ] **Helper Functions**:
-  ```cpp
-  void assert_event_at_time(const std::vector<VoiceTriggerEvent>& events,
-                            float expected_time, float tolerance = 0.001f);
-
-  void assert_event_count(const std::vector<VoiceTriggerEvent>& events,
-                          int expected_count);
-
-  TrackerScore create_test_score(const std::vector<float>& trigger_times,
-                                  float bpm = 120.0f);
-  ```
-
-**Coverage Target**: 95%+ for `src/audio/tracker.cc`.
-
----
-
-### Task #51.4: Integration with Build
-**Objective**: Wire up tests to CMake and ensure they run automatically.
-
-**Implementation Steps**:
-- [ ] **Update `src/CMakeLists.txt`**:
-  - Add `mock_audio_backend.cc` to test-only sources (under `DEMO_BUILD_TESTS`)
-  - Link `test_tracker` executable with audio subsystem library
-
-- [ ] **Add test to CTest**:
-  ```cmake
-  if(DEMO_BUILD_TESTS)
-    add_executable(test_tracker tests/test_tracker.cc audio/mock_audio_backend.cc)
-    target_link_libraries(test_tracker PRIVATE audio_lib util_lib)
-    add_test(NAME TrackerTest COMMAND test_tracker)
-  endif()
-  ```
-
-- [ ] **Verify in CI**:
-  - Run `cmake --build build && cd build && ctest`
-  - Ensure `test_tracker` runs and passes
-  - Check coverage report includes tracker.cc
-
-**Validation**: `ctest` output shows `TrackerTest: PASSED`.
-
----
-
-## Implementation Layout Summary
-
-### New Files
-```
-src/audio/audio_backend.h          # Interface definition
-src/audio/miniaudio_backend.h      # Production backend (header)
-src/audio/miniaudio_backend.cc     # Production backend (impl)
-src/audio/mock_audio_backend.h     # Test backend (header, !STRIP_ALL)
-src/audio/mock_audio_backend.cc    # Test backend (impl, !STRIP_ALL)
-src/tests/test_tracker.cc           # Comprehensive tracker tests
-```
-
-### Modified Files
-```
-src/audio/audio.cc                  # Backend abstraction layer
-src/audio/synth.cc                  # Add event hooks
-src/CMakeLists.txt                  # Add new files and tests
-```
-
-### File Structure
-```
-src/audio/
-├── audio_backend.h           [NEW] Interface (50 lines)
-├── miniaudio_backend.h       [NEW] Header (30 lines)
-├── miniaudio_backend.cc      [NEW] Production impl (~100 lines, moved from audio.cc)
-├── mock_audio_backend.h      [NEW] Test header (60 lines)
-├── mock_audio_backend.cc     [NEW] Test impl (~120 lines)
-├── audio.h                   [MODIFIED] Add backend setter
-├── audio.cc                  [MODIFIED] Use backend abstraction (~30 lines changed)
-├── synth.cc                  [MODIFIED] Add event hook (~10 lines)
-└── tracker.cc                [NO CHANGE]
-
-src/tests/
-└── test_tracker.cc           [NEW] Test suite (~400 lines)
-```
-
-### Code Organization Principles
-1. **Zero Size Impact**: All test infrastructure under `#if !defined(STRIP_ALL)`
-2. **Backward Compatible**: Production path unchanged, existing tests pass
-3. **Clean Separation**: Interface-based design, easy to add more backends later
-4. **Testable**: Mock backend has minimal dependencies (no hardware/threads)
-
----
-
 ## Future Goals
\ No newline at end of file
diff --git a/doc/COMPLETED.md b/doc/COMPLETED.md
new file mode 100644
index 0000000..168d378
--- /dev/null
+++ b/doc/COMPLETED.md
@@ -0,0 +1,270 @@
+# Completed Tasks
+
+This file tracks recently completed tasks, organized by completion date.
+
+## Recently Completed (February 7, 2026)
+
+- [x] **Audio Peak Measurement & Test Coverage Improvements** (February 7, 2026)
+    - [x] **Real-Time Peak Fix**: Fixed critical audio-visual sync bug where visual effects triggered ~400ms before audio was heard
+      - Root cause: Peak measured at ring buffer write time (synth_render) instead of playback time (audio callback)
+      - Solution: Added `get_realtime_peak()` to AudioBackend interface, implemented in MiniaudioBackend audio_callback
+      - Used exponential averaging: instant attack, 0.7 decay rate (1-second fade time)
+    - [x] **Peak Decay Optimization**: Fixed "test_demo just flashing" issue
+      - Old: 0.95 decay = 5.76 second fade (screen stayed white)
+      - New: 0.7 decay = 1.15 second fade (proper flash with smooth decay)
+    - [x] **SilentBackend Creation**: Test-only backend for audio.cc testing without hardware
+      - Created src/audio/backend/silent_backend.{h,cc}
+      - 7 comprehensive tests: lifecycle, peak control, tracking, playback time, buffer management
+      - Significantly improved audio.cc test coverage
+    - [x] **Backend Reorganization**: Moved all backends to src/audio/backend/ subdirectory
+      - Cleaner code organization, updated all includes and CMake paths
+    - [x] **Dead Code Removal**: Removed unused `register_spec_asset()` function
+    - **Result**: All 28 tests passing, perfect audio-visual sync, improved test coverage
+
+## Recently Completed (February 6, 2026)
+
+- [x] **Critical Shader Bug Fixes & Test Infrastructure** (February 6, 2026)
+    - [x] **Shader Validation Errors**: Fixed three critical WGSL bugs causing demo64k and test_3d_render crashes:
+      - Removed dead code using non-existent `inverse()` function in renderer_3d.wgsl
+      - Fixed `get_normal_basic()` signature mismatch in sdf_utils.wgsl and lighting.wgsl (obj_type: f32 → obj_params: vec4<f32>)
+      - Fixed scene_query_linear.wgsl incorrectly declaring binding 2 (BVH buffer) - was identical to BVH version due to copy-paste error
+    - [x] **Root Cause**: Linear shader mode expected no binding 2, but shader declared it, causing pipeline/shader mismatch
+    - [x] **New Test Coverage**: Created `test_shader_compilation.cc` that compiles all production shaders through WebGPU, tests both BVH and Linear composition modes, validates WGSL syntax/bindings/types
+    - [x] **Test Gap Analysis**: Existing test_shader_assets only checked keywords, not actual compilation. New test would have caught all three bugs.
+    - **Result**: demo64k runs without WebGPU errors, test_3d_render no longer crashes, 22/23 tests pass (FftTest unrelated), comprehensive regression prevention
+
+- [x] **Task C: Build System Optimization** (February 6, 2026)
+    - [x] **Header Split**: Refactored `asset_manager.h` into `asset_manager_dcl.h` (forward declarations), `asset_manager.h` (core API), and `asset_manager_utils.h` (typed helpers for TextureAsset/MeshAsset).
+    - [x] **Asset Dependency Tracking**: Added file-level dependencies for all assets (.wgsl shaders, .spec audio, .obj meshes). CMake now tracks 42 demo assets + 17 test assets individually.
+    - [x] **Performance Impact**: Editing TextureAsset/MeshAsset helpers: 4.82s → 2.01s (58% faster). Editing shaders now triggers correct rebuild (was broken before - stale code bug fixed).
+    - [x] **Developer Workflow**: No more manual `touch demo_assets.txt` workaround needed. Edit any asset file and rebuild works correctly.
+    - **Result**: Critical correctness bug fixed (shader changes were not triggering rebuilds). Build system now has proper dependency tracking for all source and asset files.
+
+## Recently Completed (February 5, 2026)
+
+- [x] **Audio Lifecycle Refactor (Task #56)**:
+    - [x] **Phase 1: Design & Prototype**: Created `AudioEngine` class and `SpectrogramResourceManager` to manage audio subsystem initialization and resource loading with lazy loading strategy.
+    - [x] **Phase 2: Test Migration**: Migrated all tracker-related tests (`test_tracker.cc`, `test_tracker_timing.cc`, `test_variable_tempo.cc`, `test_wav_dump.cc`) to use AudioEngine instead of direct synth/tracker initialization.
+    - [x] **Phase 3: Production Integration**: Updated `main.cc` to use AudioEngine, eliminating initialization order fragility in production code. Fixed pre-existing demo crash (procedural texture loading).
+    - [x] **Phase 4: Cleanup & Documentation**: Removed backwards compatibility (synth_init() from audio_init()), updated HOWTO.md and CONTRIBUTING.md with AudioEngine usage patterns and best practices.
+    - **Result**: Initialization order dependency eliminated. All 20 tests pass. Binary size impact <500 bytes. Demo runs successfully.
+
+- [x] **Physics & Collision (Task #49)**:
+    - [x] **CPU-Side SDF Library**: Implemented `sdSphere`, `sdBox`, `sdTorus`, `sdPlane` in `src/3d/sdf_cpu.h` using `mini_math.h`. Added `calc_normal` for numerical gradients.
+    - [x] **BVH Construction**: Implemented `BVHNode` and `BVHBuilder` in `src/3d/bvh.h/cc`. Optimized broad-phase collision queries with `BVH::Query`.
+    - [x] **Physics System**: Created `PhysicsSystem` in `src/3d/physics.h/cc` implementing semi-implicit Euler integration, broad-phase BVH culling, and narrow-phase SDF proxy probing.
+    - [x] **Integration & Visualization**: Added `velocity`, `mass`, `is_static` to `Object3D`. Integrated physics loop into `test_3d_render.cc` and added BVH wireframe visualization to `VisualDebug`.
+
+- [x] **Audio Playback Debugging & Core Audio Optimization**:
+    - [x] **Core Audio Timing Fix**: Resolved stop-and-go audio glitches caused by timing mismatch. Core Audio optimized for 44.1kHz (10ms periods), but our 32kHz system expected uniform ~13.78ms callbacks, causing resampling jitter. Fix: Added `allowNominalSampleRateChange = TRUE` to force OS-level 32kHz native and `performanceProfile = conservative` for 4096-frame buffers (128ms). Result: Stable ~128ms callbacks, <1ms jitter, zero underruns.
+    - [x] **Ring Buffer Capacity**: Increased from 200ms to 400ms (25,600 samples) to handle tempo scaling headroom. Added comprehensive bounds checking with abort() on violations.
+    - [x] **Tempo-Scaled Buffer Fill**: Fixed critical bug where buffer pre-fill didn't scale dt by tempo (`audio_render_ahead(g_music_time, dt * g_tempo_scale)`). Buffer now maintains 400ms fullness during 2.0x acceleration.
+    - [x] **Extensive Diagnostics**: Added high-resolution timing tracking (clock_gettime), callback interval measurement, buffer level monitoring, underrun detection. All under conditional compilation for future debugging.
+
+- [x] **NOTE_ Parsing Fix & Sample Caching**:
+    - [x] **Parser Bug Fix**: Fixed `is_note_name()` checking only first letter (A-G), causing ASSET_KICK_1 → A0 (27.5 Hz) false positives. Required "NOTE_" prefix to distinguish notes from assets. Updated music.track to use NOTE_E2, NOTE_G4 format.
+    - [x] **Resource Exhaustion Discovery**: Found every event created NEW spectrogram (14 unique samples → 228 registrations). MAX_SPECTROGRAMS=16 insufficient, causing spectrogram_id=-1 errors.
+    - [x] **Comprehensive Caching**: Implemented cache in `tracker_init()` pre-registering all samples. Assets: loaded once from AssetManager. Generated notes: created once, stored in persistent pool. All cached synth_ids reused.
+    - [x] **Memory Reduction**: MAX_SPECTROGRAMS reduced from 256 (temporary workaround) to 32 (2.3x headroom over 14 actual). 88% memory savings vs uncached approach.
+    - [x] **Resource Analysis Tool**: Enhanced `tracker_compiler` to report required/recommended pool sizes, cache potential, memory usage. Analysis showed 152/228 required without caching, 14 with caching.
+
+- [x] **Debug Logging Infrastructure**:
+    - [x] **Central Debug Header**: Created `src/util/debug.h` with 7 category macros (DEBUG_LOG_AUDIO, DEBUG_LOG_RING_BUFFER, DEBUG_LOG_TRACKER, DEBUG_LOG_SYNTH, DEBUG_LOG_3D, DEBUG_LOG_ASSETS, DEBUG_LOG_GPU).
+    - [x] **CMake Integration**: Added `DEMO_ENABLE_DEBUG_LOGS` option defining `DEBUG_LOG_ALL` to enable all categories. Individual categories can be enabled selectively.
+    - [x] **Source Code Conversion**: Updated miniaudio_backend.cc (timing, device config), ring_buffer.cc (underruns), tracker.cc (validation), synth.cc (parameter checks) to use category macros.
+    - [x] **Zero Runtime Cost**: Default build: macros compile to `((void)0)`. Debug build: comprehensive logging preserved for troubleshooting.
+    - [x] **Pre-Commit Policy**: Updated CONTRIBUTING.md requiring debug build verification before significant commits to ensure diagnostic code remains maintainable.
+    - [x] **Verification**: Both default and debug builds compile without errors. Audio playback works correctly in both modes.
+
+- [x] **Event-Based Tracker for Tempo Scaling**:
+    - [x] **Problem Identified**: Notes within patterns didn't accelerate with tempo changes. Pattern events were pre-composited into single spectrograms at fixed positions.
+    - [x] **Refactored Architecture**: Changed from pattern compositing to individual event triggering. Each TrackerEvent now triggers as separate voice.
+    - [x] **Dynamic Beat Calculation**: `elapsed_beats = (music_time - start_time) / beat_duration` allows notes to respect tempo scaling.
+    - [x] **ActivePattern Tracking**: Added structure to track pattern_id, start_music_time, and next_event_idx for each active pattern instance.
+    - [x] **Removed Compositing Logic**: Deleted paste_spectrogram approach that baked events at fixed frame offsets.
+    - [x] **Full Tempo Scaling**: At 2.0x tempo, both pattern triggering AND note spacing play 2x faster. At 0.5x tempo, both play 2x slower.
+    - [x] **Testing**: Updated test_tracker.cc to verify individual event triggers at specific beat times. All 17 tests pass.
+    - [x] **WAV Verification**: Confirmed with WAV dump showing 61.24s music time in 60s physical time during tempo transitions.
+
+- [x] **WAV Dump Backend for Debugging**:
+    - [x] **Implementation**: Created WavDumpBackend implementing AudioBackend interface to render audio offline to .wav files.
+    - [x] **Command-Line Option**: Added `--dump_wav output.wav` flag to enable offline rendering instead of live playback.
+    - [x] **Stereo Bug Fix**: Fixed critical mono/stereo mismatch. Synth outputs STEREO (interleaved L/R) but initial implementation wrote MONO.
+    - [x] **Correct Allocation**: Changed to allocate `frames * 2` samples and write stereo format (num_channels = 2).
+    - [x] **Format Matching**: WAV header now correctly specifies 16-bit PCM, stereo, 32kHz (matches live audio exactly).
+    - [x] **Regression Test**: Added test_wav_dump.cc with critical assertion `assert(header.num_channels == 2)` to prevent future mismatches.
+    - [x] **Tempo Simulation**: WAV dump backend simulates tempo scaling matching main.cc logic for accurate offline rendering.
+    - [x] **All Tests Pass**: 17/17 tests pass including WAV format verification.
+
+- [x] **Variable Tempo System**:
+    - [x] **Music Time Abstraction**: Implemented unified music time in `main.cc` that advances at `tempo_scale` rate, decoupling from physical time.
+    - [x] **Tempo Control**: Added `g_tempo_scale` (default 1.0) allowing future dynamic tempo changes without pitch shifting.
+    - [x] **Reset Tricks**: Comprehensive tests verify 2x speed-up and 2x slow-down reset techniques work correctly.
+    - [x] **Test Suite**: Created `test_variable_tempo.cc` with 6 test scenarios: basic scaling, speed-up/slow-down resets, pattern density swap, continuous acceleration, oscillating tempo.
+    - [x] **Perfect Verification**: All tests pass, confirming music_time advances correctly at variable rates (e.g., 2.0x tempo → 2x faster triggering).
+    - [x] **Zero Pitch Shift**: Spectrograms unchanged, only trigger timing affected (as designed).
+    - [x] **Documentation**: Created `ANALYSIS_VARIABLE_TEMPO_V2.md` explaining simplified trigger-timing approach.
+
+- [x] **Task #51.3 & #51.4: Tracker Test Suite & Build Integration**:
+    - [x] **Comprehensive Test Suite**: Created `test_tracker_timing.cc` with 7 test scenarios using MockAudioBackend.
+    - [x] **Simultaneous Trigger Verification**: Confirmed multiple patterns at same time trigger with **0.000ms delta** (perfect sync).
+    - [x] **Test Coverage**: Basic recording, progressive triggering, simultaneous triggers, monotonicity, seek/fast-forward, timestamp clustering, render integration.
+    - [x] **Real Music Data**: Tests use generated tracker music data for realistic validation.
+    - [x] **Build Integration**: Added to CMake with proper dependencies on generated music data.
+    - [x] **All Tests Pass**: 15/15 tests passing (100% success rate).
+
+- [x] **Task #51.2: Mock Audio Backend**:
+    - [x] **Event Recording**: Created `VoiceTriggerEvent` structure to capture timestamp, spectrogram_id, volume, and pan.
+    - [x] **MockAudioBackend Class**: Implemented test-only backend with event recording and time tracking capabilities.
+    - [x] **Time Management**: Added `advance_time()`, `set_time()`, and `get_current_time()` for deterministic testing.
+    - [x] **Frame Rendering Hook**: Implemented `on_frames_rendered()` to automatically update time based on audio frames (32kHz).
+    - [x] **Synth Integration**: Verified mock backend correctly captures voice triggers from synth engine.
+    - [x] **Comprehensive Tests**: Created `test_mock_backend.cc` with 6 test scenarios covering all mock functionality.
+    - [x] **Build Integration**: Added mock backend to test builds, all 14 tests pass.
+
+- [x] **Task #51.1: Audio Backend Abstraction**:
+    - [x] **Interface Created**: Defined `AudioBackend` interface in `src/audio/audio_backend.h` with hooks for voice triggering and frame rendering.
+    - [x] **Production Backend**: Moved miniaudio implementation from `audio.cc` to `MiniaudioBackend` class, maintaining backward compatibility.
+    - [x] **Audio Refactoring**: Updated `audio.cc` to use backend abstraction with automatic fallback to `MiniaudioBackend`.
+    - [x] **Event Hooks**: Added time tracking to `synth.cc` with `on_voice_triggered()` callbacks (guarded by `!STRIP_ALL`).
+    - [x] **Verification Test**: Created `test_audio_backend.cc` to verify backend injection and event recording work correctly.
+    - [x] **Build Integration**: Updated CMakeLists.txt to include new backend files and link audio tests properly with util/procedural dependencies.
+    - [x] **Zero Size Impact**: All test infrastructure under `#if !defined(STRIP_ALL)`, production path unchanged.
+
+- [x] **Task #50: WGSL Modularization**:
+    - [x] **Recursive Composition**: Updated `ShaderComposer` to support recursive `#include "snippet_name"` directives with cycle detection.
+    - [x] **Granular SDF Library**: Extracted `math/sdf_shapes.wgsl`, `math/sdf_utils.wgsl`, `render/shadows.wgsl`, `render/scene_query.wgsl`, and `render/lighting_utils.wgsl`.
+    - [x] **Pipeline Update**: Refactored `Renderer3D` and `renderer_3d.wgsl` to use the new modular system, reducing C++-side dependency management.
+    - [x] **Platform Fix**: Resolved `WGPUShaderSourceWGSL` usage on macOS to ensure compatibility with composed shader strings.
+
+- [x] **Task #48: Improve Audio Coverage**:
+    - [x] **New Tests**: Added `test_dct` (100% coverage for transforms) and `test_audio_gen` (94% coverage for procedural audio).
+    - [x] **Enhanced Tests**: Updated `test_synth` to cover rendering loop, double-buffering, and resource exhaustion.
+    - [x] **Coverage Boost**: Increased `src/audio/` coverage from ~42% to 93%.
+- [x] **Task #47: Improve Asset Manager Coverage**:
+    - [x] **New Tests**: Added tests for unknown procedural functions, generation failures, and edge cases in `src/tests/test_assets.cc`.
+    - [x] **Tooling Update**: Downgraded `asset_packer` validation error to warning to allow testing invalid assets.
+    - [x] **Coverage Boost**: Increased `src/util/asset_manager.cc` coverage from 71% to 88%.
+- [x] **Task #46: Enhance Coverage Script**: Updated `scripts/gen_coverage_report.sh` to accept an optional directory argument for targeted coverage reports (e.g., `src/procedural`).
+- [x] **Task #45: Improve Procedural Generation Coverage**:
+    - [x] **Unit Tests:** Implemented comprehensive tests for `gen_perlin`, `make_periodic`, and default parameter handling in `src/tests/test_procedural.cc`.
+    - [x] **Coverage Boost:** Increased `src/procedural/generator.cc` coverage from 38% to 96%.
+- [x] **Task #44: Developer Tooling (Coverage)**:
+    - [x] **Implement Code Coverage:** Added `DEMO_ENABLE_COVERAGE` CMake option and created `scripts/gen_coverage_report.sh` to generate HTML coverage reports using `lcov` on macOS.
+    - [x] **Documentation:** Updated `doc/HOWTO.md` with usage instructions.
+- [x] **Skybox & Two-pass Rendering Stability**:
+    - [x] **Fixed Two-pass Rendering:** Implemented mandatory clear operations for color and depth when the skybox is absent, preventing black screens and depth validation errors.
+    - [x] **Implemented Rotating Skybox:** Added `inv_view_proj` to `GlobalUniforms` and updated the skybox shader to perform world-space ray unprojection (`inv_view_proj`), enabling correct rotation with the camera.
+    - [x] **Enhanced Procedural Noise:** Implemented a multi-octave Value Noise generator for higher-quality skybox textures.
+    - [x] **Scene Integrity:** Restored proper object indexing and removed redundant geometry, ensuring the floor grid and objects render correctly.
+
+- [x] **Task #57: Interactive Timeline Editor (Phase 1 Complete)** 🎉
+    - [x] **Core Parser & Renderer**: Implemented demo.seq parser with BPM, beat notation, priority modifiers. Gantt-style timeline rendering with dynamic sequence/effect positioning.
+    - [x] **Drag & Drop**: Sequences and effects draggable along timeline with proper offset calculation. Fixed critical e.target vs e.currentTarget bug preventing erratic jumping.
+    - [x] **Resize Handles**: Left/right handles on selected effects. Allow negative relative times (effects extend before sequence start).
+    - [x] **Snap-to-Beat**: Checkbox toggle with beat markers. Automatic snapping when dragging in beat mode.
+    - [x] **Properties Panel**: Floating, collapsible panel with auto-apply (no Apply button). Properties update on input change.
+    - [x] **Stack-Order Priority**: Up/Down buttons to reorder effects within sequence. Stack position determines rendering priority (higher index = rendered later = on top).
+    - [x] **Priority Modifiers**: Toggle button for "Same as Above" (= modifier) vs "Increment" (+ modifier). Visual feedback in button text.
+    - [x] **Diagonal Scroll**: Mouse wheel navigation with 10% viewport slack. Smooth following to time-ordered sequence cascade. Flash animation on active sequence change.
+    - [x] **Dynamic Bounds**: Sequence visual bounds calculated from min/max effect times. Cumulative Y positioning prevents overlap.
+    - [x] **Waveform Visualization**: Load WAV files (via Web Audio API) and display waveform above timeline. Scales with zoom (pixels per second). Documented integration with `--dump_wav` flag.
+    - [x] **UI Polish**: Hoverable sequence names (large centered text, fades on hover). No scrollbars (hidden CSS). Clean, minimal interface.
+    - [x] **File I/O**: Load/save demo.seq files with proper serialization. BPM parsing and display. Re-order sequences by time.
+    - [x] **Delete & Add**: Delete sequences/effects. Add new sequences (at time 0, priority 0).
+    - **Result**: Fully functional timeline editor ready for production use. Phase 1.1 complete (basic editing, snap-to-beat, waveform). Phase 1.2 (Add Effect button) and Phase 2.5 (music.track visualization) documented in ROADMAP.md.
+    - **Files**: `tools/timeline_editor/index.html` (~1200 lines), `README.md`, `ROADMAP.md` (3 phases, 117-161 hour estimate).
+
+---
+
+## Milestones from PROJECT_CONTEXT.md
+
+#### Milestone: Audio Peak Measurement & Test Coverage (February 7, 2026) ✅
+- **Real-Time Audio Peak Fix**: Resolved critical audio-visual synchronization bug where visual effects triggered ~400ms before audio was heard. Root cause: peak measurement occurred at ring buffer write time (synth_render) instead of playback time (audio callback). Solution: Added `get_realtime_peak()` to AudioBackend interface, implemented real-time peak tracking in MiniaudioBackend audio_callback using exponential averaging (instant attack, 0.7 decay rate for 1-second fade). Updated main.cc and test_demo.cc to use `audio_get_realtime_peak()` API. Result: Perfect audio-visual synchronization, visual effects trigger precisely when audio is heard.
+
+- **Peak Decay Optimization**: Fixed "test_demo just flashing" issue caused by slow decay rate. Old: 0.95 decay = 5.76 second fade (screen stayed white constantly). New: 0.7 decay = 1.15 second fade (proper flash with smooth decay). Mathematical verification: At 128ms audio callbacks, 0.7^9 ≈ 0.1 after ~1 second. Result: Responsive visual effects with proper attack/decay curves.
+
+- **SilentBackend Test Infrastructure**: Created test-only audio backend for comprehensive audio.cc testing without hardware. Features: controllable peak for edge cases, tracks frames rendered and voice triggers, pure inspection (no audio output). Created 7 comprehensive tests covering lifecycle, peak control, frame/voice tracking, playback time, buffer management, audio_update safety. Result: Significantly improved audio.cc test coverage (audio_get_playback_time, audio_render_ahead, audio_update all tested).
+
+- **Backend Reorganization**: Moved all audio backend implementations to src/audio/backend/ subdirectory for cleaner code organization. Moved: miniaudio_backend, mock_audio_backend, wav_dump_backend, jittered_audio_backend, silent_backend (5 backends total). Updated all #include paths in backend files and tests, updated CMakeLists.txt paths. Kept audio_backend.h (interface) in src/audio/. Result: Improved maintainability, zero functionality regressions, all 28 tests pass.
+
+- **Dead Code Removal**: Removed unused `register_spec_asset()` function from audio.{h,cc}. Function was never called anywhere in codebase. Removed declaration and implementation (lines 43-58). Result: Cleaner codebase, reduced maintenance burden.
+
+- **Test Coverage Achievement**: All 28 tests passing (100% success rate). New SilentBackend tests provide comprehensive coverage for audio subsystem. No regressions in existing tests. Build clean with no errors or warnings. Files modified: 15 changed, 3 new files, 8 files reorganized. Commit a6a7bf0 ready for push.
+
+#### Milestone: test_demo - Audio/Visual Sync Debug Tool (February 7, 2026) 🎯
+- **Standalone Debug Tool**: Created minimal test executable for debugging audio/visual synchronization and variable tempo system without full demo complexity. **Core Features**: Simple drum beat (kick-snare) with crash landmarks at bars 3 and 7, NOTE_A4 (440 Hz) reference tone at start of each bar for testing, screen flash effect synchronized to audio peaks, 16 second duration (8 bars at 120 BPM). **Variable Tempo Mode**: `--tempo` flag enables alternating tempo scaling (even bars: 1.0x → 1.5x acceleration, odd bars: 1.0x → 0.66x deceleration) to test music time vs physical time system. **Peak Logging**: `--log-peaks FILE` exports audio peak data for gnuplot visualization, `--log-peaks-fine` adds millisecond-resolution per-frame logging (~960 samples vs 32 beat-aligned samples). **Command-Line Options**: `--help` shows usage, `--fullscreen` for fullscreen mode, `--resolution WxH` for custom window size. **Error Handling**: Unknown options print error message and help text before exiting with status 1. **File Structure**: `src/test_demo.cc` (main executable, ~220 lines), `assets/test_demo.track` (drum patterns with NOTE_A4), `assets/test_demo.seq` (visual timeline), `test_demo_README.md` (comprehensive documentation). **Peak Log Format**: Beat-aligned mode logs at beat boundaries (32 samples), fine-grained mode logs every frame with beat_number column for correlation. **Build Integration**: CMake target with timeline/music generation, proper dependencies, size optimizations. **Use Cases**: Verify millisecond-precision sync, detect timing jitter, analyze tempo scaling effects, debug flash-to-audio alignment. **Impact**: Provides isolated testing environment for sync verification without demo64k complexity, enables data-driven analysis via exported logs.
+
+- **CMake Configuration Summary**: Added formatted configuration summary to CMake output displaying all build options (DEMO_SIZE_OPT, DEMO_STRIP_ALL, DEMO_BUILD_TESTS, etc.) with ON/OFF status, build type, and compiler information. Appears automatically at end of `cmake` command for easy verification of enabled options.
+
+- **Code Quality Fixes**: Replaced deprecated `std::sprintf` with `std::snprintf` in `asset_packer.cc` to eliminate compiler warnings and add buffer overflow protection during OBJ mesh vertex key generation.
+
+#### Milestone: Audio Playback Stability & Debug Infrastructure (February 4, 2026)
+- **Core Audio Backend Optimization**: Resolved critical audio playback issues (stop-and-go, glitches, eventual silence) caused by timing mismatches in miniaudio's Core Audio backend. Root cause: Core Audio optimized for 44.1kHz with 10ms periods, but our 32kHz system expected uniform ~13.78ms callbacks, causing resampling jitter. Fix: Added `allowNominalSampleRateChange = TRUE` to force OS-level 32kHz native and `performanceProfile = conservative` for larger buffers (4096 frames = 128ms). Result: Stable ~128ms callbacks with <1ms jitter, zero underruns during variable tempo (1.0x → 2.0x).
+
+- **Ring Buffer Capacity Tuning**: Increased ring buffer from 200ms to 400ms (25,600 samples) to handle tempo scaling headroom. Added comprehensive bounds checking with abort() on violations to catch buffer corruption early. Fixed critical bug: tempo-scaled rendering wasn't scaling dt when pre-filling buffer (`audio_render_ahead(g_music_time, dt * g_tempo_scale)`). Buffer now maintains 400ms fullness throughout playback, including 2.0x tempo acceleration.
+
+- **NOTE_ Parsing Bug Fix & Sample Caching**: Fixed critical `is_note_name()` bug that only checked first letter (A-G), causing ASSET_KICK_1 to be misidentified as A0 (27.5 Hz). Solution: Required "NOTE_" prefix (NOTE_E2, NOTE_G4) to distinguish from ASSET_* samples. Discovered resource exhaustion: every note event created NEW spectrogram (14 unique samples → 228 registrations). Implemented comprehensive caching in `tracker_init()`: pre-register all asset samples (loaded once from AssetManager) and generated notes (created once, stored in persistent pool). Result: 256 → 32 MAX_SPECTROGRAMS (88% memory reduction), zero spectrogram_id=-1 errors, perfect audio playback.
+
+- **Debug Logging Infrastructure**: Created systematic debug logging system (`src/util/debug.h`) with 7 category macros (DEBUG_LOG_AUDIO, DEBUG_LOG_RING_BUFFER, DEBUG_LOG_TRACKER, DEBUG_LOG_SYNTH, DEBUG_LOG_3D, DEBUG_LOG_ASSETS, DEBUG_LOG_GPU). Added `DEMO_ENABLE_DEBUG_LOGS` CMake option that defines `DEBUG_LOG_ALL` to enable all categories. Converted all diagnostic code in audio subsystem (miniaudio callbacks, ring buffer tracking, tracker validation, synth parameter checks) to use category macros. Default build: macros compile to `((void)0)` for zero runtime cost. Debug build: comprehensive logging (timing, buffer state, caching, validation). Updated `CONTRIBUTING.md` with pre-commit policy requiring debug build verification to ensure diagnostic code remains maintainable.
+
+- **Resource Analysis Tool**: Enhanced `tracker_compiler` to report required vs recommended pool sizes, cache potential, and memory usage. Analysis showed 152 required / 228 recommended spectrograms without caching, only 14 unique samples with caching. Tool now generates actionable optimization recommendations during music data compilation.
+
+#### Milestone: Audio System Robustness & Variable Tempo (February 4, 2026)
+- **Event-Based Tracker for Tempo Scaling**: Refactored tracker system from pattern compositing to individual event triggering. Previously, pattern events were baked into single spectrograms at fixed positions, so tempo scaling only affected pattern trigger timing. Now each TrackerEvent triggers as a separate voice with timing calculated dynamically based on music_time. Result: Notes within patterns correctly accelerate/decelerate with tempo changes. At 2.0x tempo, both pattern triggering AND note spacing play 2x faster. Verified with WAV dump showing 61.24s music time in 60s physical time during tempo transitions. All 17 tests pass (100%).
+
+- **WAV Dump Backend for Debugging**: Added `WavDumpBackend` implementing `AudioBackend` interface to render audio offline to .wav files instead of playing on device. Enabled via `--dump_wav output.wav` flag. Critical bug fix: Synth outputs STEREO (interleaved L/R) but initial implementation wrote MONO, causing severe distortion. Fixed by allocating `frames * 2` samples and writing stereo format (num_channels = 2). Added regression test (`test_wav_dump.cc`) with critical assertion `assert(header.num_channels == 2)` to prevent future mono/stereo mismatches. WAV format now matches live audio exactly: 16-bit PCM, stereo, 32kHz.
+
+- **Variable Tempo System**: Implemented unified music time abstraction in `main.cc` that decouples tracker timing from physical time. Music time advances at configurable `tempo_scale` rate (default 1.0), enabling dynamic tempo changes without pitch shifting. Created comprehensive test suite (`test_variable_tempo.cc`) verifying 2x speed-up and 2x slow-down "reset tricks" work perfectly. All 6 test scenarios pass with mathematical precision. System ready for expressive tempo control in demo with zero size impact.
+
+- **Task #51: Tracker Timing Verification System**: Created robust audio testing infrastructure with mock backend abstraction. Implemented `AudioBackend` interface separating synthesis from output, `MiniaudioBackend` for production, and `MockAudioBackend` for testing. Added event recording with precise timestamp tracking (32kHz sample rate). Created comprehensive test suite (`test_tracker_timing.cc`) that **verified simultaneous pattern triggers have 0.000ms delta** (perfect synchronization). All test infrastructure under `#if !defined(STRIP_ALL)` for zero production impact. 17/17 tests passing.
+
+#### Milestone: Audio Lifecycle Refactor (February 5, 2026)
+- **Task #56: AudioEngine Implementation**: Completed 4-phase refactor to eliminate fragile initialization order dependency between synth and tracker. **Phase 1 (Design & Prototype)**: Created `AudioEngine` class and `SpectrogramResourceManager` for unified lifecycle management with lazy loading. **Phase 2 (Test Migration)**: Migrated all tracker tests to use AudioEngine (test_tracker.cc, test_tracker_timing.cc, test_variable_tempo.cc, test_wav_dump.cc). **Phase 3 (Production Integration)**: Updated main.cc to use AudioEngine, fixed pre-existing procedural texture crash (flash_cube_effect.cc, hybrid_3d_effect.cc). **Phase 4 (Cleanup & Documentation)**: Removed backwards compatibility (synth_init from audio_init), updated HOWTO.md and CONTRIBUTING.md with usage patterns. Result: All 20 tests pass, binary size impact <500 bytes, initialization fragility eliminated.
+
+#### Milestone: Build System Optimization (February 6, 2026)
+- **Task C: CMake Dependency Graph Optimization**: Resolved critical build correctness bugs and improved developer iteration speed. **Header Split**: Refactored monolithic `asset_manager.h` (61 lines) into three focused headers: `asset_manager_dcl.h` (forward declarations for AssetId), `asset_manager.h` (core GetAsset/DropAsset API), and `asset_manager_utils.h` (typed helpers for TextureAsset/MeshAsset). Updated 17 source files to use appropriate headers. **Asset Dependency Tracking**: Implemented file-level dependency tracking for all 42 demo assets and 17 test assets. CMake now correctly tracks individual `.wgsl` shaders, `.spec` audio files, and `.obj` mesh files. **Critical Bug Fixed**: Changing shader files was NOT triggering asset regeneration, resulting in stale code in binaries. Developers had to manually `touch demo_assets.txt` as workaround. Now works correctly. **Performance**: Editing TextureAsset/MeshAsset helpers improved from 4.82s to 2.01s (58% faster). Shader edits now trigger correct 3.5s rebuild (was 0.28s with no rebuild - incorrect). **Implementation**: Added `parse_asset_list()` CMake function that parses `demo_assets.txt` format (`ASSET_NAME, COMPRESSION, FILENAME, DESC`) and extracts individual file paths for dependency tracking. All 20 tests pass. Zero functionality regressions.
+
+#### Milestone: Critical Shader Stability & Test Infrastructure (February 6, 2026) ✅
+- **Shader Crash Resolution**: Fixed three critical WGSL validation errors causing demo64k and test_3d_render to crash on startup. **Bug 1 (renderer_3d.wgsl)**: Removed dead code using non-existent `inverse()` function - WGSL doesn't provide matrix inverse, and validator checks all code paths even unreachable ones. Also removed reference to undefined `in.normal` vertex input. **Bug 2 (sdf_utils.wgsl, lighting.wgsl)**: Fixed `get_normal_basic()` function signature mismatch - changed from `obj_type: f32` to `obj_params: vec4<f32>` to match `get_dist()` calls. **Bug 3 (scene_query_linear.wgsl - ROOT CAUSE)**: Fixed linear scene query mode incorrectly declaring binding 2 (BVH storage buffer). Linear version was identical to BVH version due to copy-paste error. Replaced BVH traversal with proper linear object iteration loop. **Impact**: When `use_bvh=false`, pipeline created without binding 2, but shader expected it → validation error → crash.
+
+- **Comprehensive Shader Compilation Tests**: Created `test_shader_compilation.cc` to prevent regression. Test compiles all production shaders through WebGPU (`wgpuDeviceCreateShaderModule`), validates both BVH and Linear composition modes using `ShaderComposer`, catches WGSL syntax errors, binding mismatches, type errors, and function signature issues. **Test Gap Analysis**: Existing `test_shader_assets.cc` only checked for keywords (`@vertex`, `fn`, etc.) without actual compilation - would NOT have caught any of the three bugs. New test fills this gap with real GPU validation. **Graceful Degradation**: Test handles platforms where WebGPU device unavailable by skipping GPU tests but still validating shader composition.
+
+- **Results**: demo64k runs cleanly without WebGPU errors, test_3d_render no longer crashes, 22/23 tests pass (FftTest unrelated to this work), comprehensive regression prevention infrastructure in place. All shader composition modes (BVH/Linear) now validated in CI. Files modified: renderer_3d.wgsl, sdf_utils.wgsl, lighting.wgsl, scene_query_linear.wgsl, test_shader_compilation.cc (new), CMakeLists.txt.
+
+#### Milestone: FFT-based DCT/IDCT Complete (February 6, 2026) ✅
+- **Core FFT Implementation**: Replaced failing double-and-mirror method with Numerical Recipes reordering method for DCT via FFT. Fixed reference IDCT to use DCT-III (inverse of DCT-II, not IDCT-II). All transforms now use orthonormal normalization: sqrt(1/N) for DC term, sqrt(2/N) for AC terms. Result: Perfect round-trip accuracy for impulse signals, <5e-3 error for sinusoidals (acceptable for FFT).
+
+- **Audio Pipeline Integration**: Integrated FFT-based DCT/IDCT into audio engine (`src/audio/idct.cc`, `fdct.cc`) and both web editors (`tools/spectral_editor/dct.js`, `tools/editor/dct.js`). All synthesis paths now use fast O(N log N) FFT instead of O(N²) naive DCT. Regenerated all 14 spectrogram assets with orthonormal DCT to match new synthesis engine.
+
+- **Critical Fixes**: **(1) Normalization Mismatch**: Old non-orthonormal DCT produced 16× larger values. Solution: Regenerated all `.spec` assets with orthonormal DCT. **(2) Procedural Notes**: NOTE_* generation inaudible after normalization change. Solution: Added sqrt(DCT_SIZE/2) = 16× scaling compensation in `gen.cc`. **(3) Windowing Error**: Hamming window incorrectly applied to spectrum before IDCT (should only be used for analysis, not synthesis). Solution: Removed windowing from `synth.cc` and both editors. Result: Correct volume, no distortion, clean frequency spectrum.
+
+- **Testing & Verification**: All 23 tests pass (100% success rate). Round-trip accuracy verified (impulse at index 0: perfect). Audio playback: correct volume, no distortion. Procedural notes: audible at correct levels. Web editors: clean spectrum, no comb artifacts. WAV dumps match expected output.
+
+- **Key Technical Insights**: (1) DCT-III is the inverse of DCT-II, not IDCT-II. (2) Hamming window is ONLY for analysis (before DCT), NOT synthesis (before IDCT). (3) Orthonormal DCT produces sqrt(N/2) smaller values than non-orthonormal. (4) Reordering method is more accurate than double-and-mirror for DCT via FFT. (5) Round-trip accuracy is more important than absolute DCT accuracy.
+
+#### Milestone: Interactive Timeline Editor (February 5, 2026) 🎉
+- **Task #57 Phase 1: Production-Ready Timeline Editor**: Created fully functional web-based editor for `demo.seq` timeline files. **Core Features**: Load/save demo.seq with BPM parsing, Gantt-style visual timeline, drag & drop sequences/effects with snap-to-beat, resize handles (left/right) allowing negative relative times, stack-order based priority system (Up/Down buttons + "Same as Above" toggle), floating auto-apply properties panel, diagonal mouse wheel scroll with 10% viewport slack, dynamic sequence bounds calculation, delete/add sequences, re-order by time. **Audio Visualization**: WAV waveform display above timeline using Web Audio API, scales with zoom (pixels per second), documented integration with `--dump_wav` flag for aligning sequences with actual demo audio output. **UI Polish**: Hoverable sequence names (large centered, fades on hover), hidden scrollbars, crosshair cursor on waveform, flash animation on active sequence change, clean minimal interface. **Bug Fixes**: Resolved critical e.target vs e.currentTarget drag offset bug, fixed sequence overlap with cumulative Y positioning, corrected effect relative time calculations. **Files**: `tools/timeline_editor/index.html` (~1200 lines pure HTML/CSS/JS, no dependencies), `README.md` (usage guide, wav_dump integration docs), `ROADMAP.md` (3 phases, 117-161 hour estimate). **Next**: Phase 1.2 (Add Effect button), Phase 2.5 (music.track visualization overlay). **Impact**: Visual timeline editing now production-ready, eliminates manual text editing for sequence placement and timing adjustments.
+
+#### Other Completed Tasks
+- **Task #50: WGSL Modularization**: Updated `ShaderComposer` to support recursive `#include` directives, refactored the entire shader library into granular snippets (shapes, utils, lighting), and updated the 3D renderer to use this modular system. This resolved macOS shader compilation issues and significantly improved shader maintainability.
+- **Task #48: Improve Audio Coverage**: Achieved 93% coverage for `src/audio/` by adding dedicated tests for DCT transforms, procedural generation, and synthesis rendering.
+- **Task #47: Improve Asset Manager Coverage**: Increased `asset_manager.cc` coverage to 88% by testing runtime error paths (unknown functions, generation failure).
+- **Task #46: Enhance Coverage Script**: Updated coverage report script to support directory filtering (e.g., `./scripts/gen_coverage_report.sh src/procedural`).
+- **Task #45: Improve Procedural Generation Coverage**: Achieved 96% coverage for `src/procedural/` by implementing comprehensive tests for Perlin noise, periodic blending, and parameter handling.
+- **Task #44: Developer Tooling (Coverage)**: Added `DEMO_ENABLE_COVERAGE` CMake option and created `scripts/gen_coverage_report.sh` to generate HTML coverage reports using `lcov` on macOS.
+- **Skybox & Two-pass Rendering Stability**: Resolved "black screen" and validation errors by implementing a robust two-pass rendering architecture (Pass 1: Skybox/Clear, Pass 2: Scene Objects). Implemented a rotating skybox using world-space ray unprojection (`inv_view_proj`) and a multi-octave procedural noise generator.
+- **Task #20: Platform & Code Hygiene**: Consolidated platform-specific shims and WebGPU headers into `platform.h`. Refactored `platform_init` and `platform_poll` for better abstraction. Removed STL containers from initial hot paths (`AssetManager`, `procedural`). Full STL removal for CRT replacement is deferred to the final optimization phase.
+- **Task #26: Shader Asset Testing & Validation**: Developed comprehensive tests for `ShaderComposer` and WGSL asset loading/composition. Added a shader validation test to ensure production assets are valid.
+- **Asset Pipeline Improvement**: Created a robust `gen_spectrograms.sh` script to automate the conversion of `.wav` and `.aif` files to `.spec` format, replacing the old, fragile script. Added 13 new drum and bass samples to the project.
+- **Build System Consolidation (Task #25)**: Modularized the build by creating subsystem libraries (audio, gpu, 3d, util, procedural) and implemented helper macros to reduce boilerplate in `CMakeLists.txt`. This improves build maintenance and prepares for future CRT replacement.
+- **Asset System Robustness**: Resolved "static initialization order fiasco" by wrapping the asset table in a "Construct On First Use" getter (`GetAssetRecordTable()`), ensuring assets are available during dynamic global initialization (e.g., shader strings).
+- **Shader Asset Integration (Task #24)**: Extracted all hardcoded WGSL strings into `.wgsl` assets, registered them in `demo_assets.txt`, and updated `Renderer3D`, `VisualDebug`, and `Effects` to use `GetAsset` and `ShaderComposer`.
+- **WebGPU Stabilization**: Resolved `WGPUSurface` creation failures on macOS by adding platform-specific `GLFW_EXPOSE_NATIVE_COCOA` definitions and fixed validation errors in the render pass configuration.
+- **Final Build Stripping (Task #8)**: Implemented the `STRIP_ALL` macro to remove non-essential code (CLI parsing, debug labels, iostream) and refined size optimization flags (`-dead_strip`) for macOS.
+- **Minimal Audio Tracker (Task 21.3)**: Finalized a pattern-based audio tracker supporting both procedural notes and asset-based spectrograms with a unified "one-voice-per-pattern" pasting strategy.
+- **WGSL Library (Task 21.1)**: Implemented `ShaderComposer` for modular WGSL snippet management.
+- **Tight Ray Bounds (Task 21.2)**: Implemented local-space ray-box intersection to optimize SDF raymarching.
+- **High-DPI Fix**: Resolved viewport "squishing" via dynamic resolution uniforms and explicit viewports.
+- **Unified 3D Shadows**: Implemented robust SDF shadows across all objects using `inv_model` transforms.
+- **test_mesh tool**: Implemented a standalone `test_mesh` tool for visualizing OBJ files with debug normal display.
diff --git a/doc/CONTEXT_MAINTENANCE.md b/doc/CONTEXT_MAINTENANCE.md
new file mode 100644
index 0000000..c228cd7
--- /dev/null
+++ b/doc/CONTEXT_MAINTENANCE.md
@@ -0,0 +1,332 @@
+# Context Maintenance Guide
+
+This document describes how to maintain clean, efficient context for AI agents (Claude, Gemini) working on this project.
+
+## Philosophy
+
+**Goal**: Keep AI context focused on **actionable, current information** while archiving **historical details** for reference.
+
+**Principle**: The context should answer "What am I working on **now**?" not "What did we do 6 months ago?"
+
+---
+
+## File Organization Hierarchy
+
+### Tier 1: Critical (Always Loaded)
+These files are essential for every session and should remain lean:
+
+- **CLAUDE.md** / **GEMINI.md** - Agent configuration (keep to ~30 lines)
+- **PROJECT_CONTEXT.md** - High-level project overview (keep to ~200 lines)
+  - Current status summary (not detailed milestones)
+  - Architectural overview (brief)
+  - Next priorities (not detailed plans)
+- **TODO.md** - Active tasks only (keep to ~300 lines)
+  - Current and next tasks
+  - Brief attack plans (not 200-line implementation details)
+  - Mark completed tasks and reference COMPLETED.md
+- **README.md** - Quick start guide
+
+### Tier 2: Technical Reference (Always Loaded)
+Essential for daily work:
+
+- **doc/HOWTO.md** - Usage instructions, common commands
+- **doc/CONTRIBUTING.md** - Coding guidelines, commit policies
+- **doc/AI_RULES.md** - Agent-specific rules
+
+### Tier 3: Design Docs (Load On-Demand)
+Load these only when working on specific subsystems:
+
+- **doc/ASSET_SYSTEM.md** - Asset pipeline details
+- **doc/BUILD.md** - Build system details
+- **doc/3D.md** - 3D rendering architecture
+- **doc/SPEC_EDITOR.md** - Spectral editor design
+- **doc/TRACKER.md** - Audio tracker system
+- **doc/PROCEDURAL.md** - Procedural generation
+- **doc/ANALYSIS_VARIABLE_TEMPO_V2.md** - Tempo system analysis
+
+### Tier 4: Historical Archive (Load Rarely)
+Reference material for understanding past decisions:
+
+- **doc/COMPLETED.md** - Detailed completion history
+- **doc/HANDOFF*.md** - Agent handoff documents
+- **doc/*_ANALYSIS.md** - Technical investigations
+- **doc/*_SUMMARY.md** - Task/feature summaries
+
+---
+
+## Maintenance Schedule
+
+### After Major Milestone (Immediately)
+**Trigger**: Completed a multi-week task or significant feature
+
+**Actions**:
+1. ✅ Move detailed completion notes from TODO.md to COMPLETED.md
+2. ✅ Update PROJECT_CONTEXT.md "Current Status" (remove old "Recently Completed")
+3. ✅ Archive detailed implementation plans from TODO.md
+4. ✅ Update "Next Up" section in PROJECT_CONTEXT.md
+
+**Time**: 10-15 minutes
+
+### Monthly Review (1st of Month)
+**Trigger**: Calendar-based
+
+**Actions**:
+1. ✅ Archive tasks completed >30 days ago from TODO.md → COMPLETED.md
+2. ✅ Review PROJECT_CONTEXT.md for outdated "Recently Completed" items
+3. ✅ Check for temporary analysis docs in root directory → move to doc/
+4. ✅ Verify CLAUDE.md/GEMINI.md only load Tier 1-2 files by default
+
+**Time**: 15-20 minutes
+
+### On-Demand (When Context Feels Slow)
+**Trigger**: AI responses taking longer, context bloat suspected
+
+**Actions**:
+1. ✅ Run token analysis: `wc -w CLAUDE.md PROJECT_CONTEXT.md TODO.md doc/*.md | sort -rn`
+2. ✅ Check for oversized files (>500 lines in Tier 1-2)
+3. ✅ Aggressively archive verbose sections
+4. ✅ Consider splitting large design docs
+
+**Time**: 30-45 minutes
+
+---
+
+## What to Archive
+
+### ✅ ALWAYS Archive:
+- Detailed implementation plans for completed tasks (>100 lines)
+- Verbose milestone descriptions with step-by-step breakdowns
+- Analysis documents for resolved issues
+- Deprecated design docs (keep note in main doc pointing to archive)
+- Old handoff documents (>1 month old)
+
+### ⚠️ SUMMARIZE Before Archiving:
+- Major architectural decisions (keep 2-3 line summary)
+- Critical bug resolutions (keep root cause + solution)
+- Performance optimizations (keep before/after metrics)
+
+### ❌ NEVER Archive:
+- Current task priorities
+- Essential coding guidelines (CONTRIBUTING.md, AI_RULES.md)
+- Quick-start instructions (README.md, HOWTO.md)
+- Active architectural constraints
+
+---
+
+## Red Flags (Time to Clean Up)
+
+### File Size Red Flags:
+- PROJECT_CONTEXT.md >500 lines
+- TODO.md >800 lines
+- Any Tier 1-2 file >1000 lines
+
+### Content Red Flags:
+- "Recently Completed" section >100 lines in PROJECT_CONTEXT.md
+- TODO.md contains tasks marked `[DONE]` for >30 days
+- Detailed implementation plans (>50 lines) for completed tasks in TODO.md
+- Analysis documents in root directory (should be in doc/)
+
+### Context Red Flags:
+- AI responses noticeably slower
+- Token usage >50K on initial load
+- Repeated "let me read..." for basic project info
+
+---
+
+## Archiving Workflow
+
+### 1. Identify Content to Archive
+```bash
+# Find large files
+find . -name "*.md" -type f -exec wc -l {} + | sort -rn | head -10
+
+# Find old completed tasks in TODO.md
+grep -n "\[x\].*202[0-9]-[0-9][0-9]-[0-9][0-9]" TODO.md
+```
+
+### 2. Move to COMPLETED.md
+```markdown
+## Recently Completed (YYYY-MM-DD)
+
+- [x] **Task Name**: Brief summary
+  - Key points (2-3 bullets max)
+  - Result: One-line outcome
+```
+
+### 3. Update Source Files
+- Replace verbose sections with: `**Note**: Detailed history in doc/COMPLETED.md`
+- Update cross-references
+- Keep high-level summaries
+
+### 4. Update Agent Configs
+Update CLAUDE.md and GEMINI.md if files moved:
+```markdown
+# Design docs (load on-demand when needed)
+# Use: "read @doc/COMPLETED.md" for detailed history
+```
+
+### 5. Verify
+```bash
+# Check context size
+cat CLAUDE.md PROJECT_CONTEXT.md TODO.md | wc -w
+# Target: <10,000 words for Tier 1 files
+
+# Verify no broken references
+grep -r "@doc/" CLAUDE.md GEMINI.md PROJECT_CONTEXT.md TODO.md
+```
+
+---
+
+## Common Mistakes to Avoid
+
+### ❌ Don't:
+- Archive current priorities or active tasks
+- Remove essential architectural context
+- Break cross-references without updating
+- Move files without updating CLAUDE.md/GEMINI.md
+- Delete content (archive instead)
+- Keep outdated "Recently Completed" sections >3 months
+
+### ✅ Do:
+- Preserve key technical insights (in summaries)
+- Update file references after moving docs
+- Keep audit trail (note where content moved)
+- Maintain searchable keywords (for finding archives)
+- Regular small cleanups (better than big purges)
+
+---
+
+## Quick Reference: File Moves
+
+When moving files from root to doc/:
+
+```bash
+# 1. Move file (use git mv if tracked)
+git mv OLDFILE.md doc/OLDFILE.md
+
+# 2. Update CLAUDE.md
+# Change: @OLDFILE.md
+# To: # Use: "read @doc/OLDFILE.md" when needed
+
+# 3. Update GEMINI.md (same as above)
+
+# 4. Update cross-references in other files
+grep -r "OLDFILE.md" *.md doc/*.md
+
+# 5. Commit with clear message
+git commit -m "docs: Move OLDFILE.md to doc/ for better organization"
+```
+
+---
+
+## Context Size Targets
+
+### Optimal Targets:
+- **CLAUDE.md**: <30 lines
+- **PROJECT_CONTEXT.md**: <300 lines
+- **TODO.md**: <500 lines
+- **Total Tier 1-2**: <10,000 words (~40K tokens)
+
+### Warning Thresholds:
+- **PROJECT_CONTEXT.md**: >500 lines
+- **TODO.md**: >800 lines
+- **Total Tier 1-2**: >15,000 words (~60K tokens)
+
+### Critical (Clean Up Immediately):
+- **PROJECT_CONTEXT.md**: >800 lines
+- **TODO.md**: >1200 lines
+- **Total Tier 1-2**: >20,000 words (~80K tokens)
+
+---
+
+## Examples
+
+### Good: Lean TODO.md Entry
+```markdown
+- [ ] **Task #72: Implement Audio Compression**
+    - Goal: Reduce .spec file sizes by 50%
+    - Approach: Quantize to uint16_t, use RLE encoding
+    - See: doc/AUDIO_COMPRESSION_PLAN.md for details
+```
+
+### Bad: Verbose TODO.md Entry
+```markdown
+- [ ] **Task #72: Implement Audio Compression**
+    - Goal: Reduce .spec file sizes by 50%
+    - Background: Currently using float32 which is wasteful...
+    - Research: Investigated 5 compression algorithms...
+    - Step 1: Create compression interface
+      - Define CompressorBase class
+      - Add compress() and decompress() methods
+      - Implement factory pattern for algorithm selection
+    - Step 2: Implement RLE compressor
+      - [100+ more lines of detailed implementation]
+```
+
+### Good: Lean PROJECT_CONTEXT.md Status
+```markdown
+### Current Status
+- Audio: Stable, 93% test coverage, variable tempo support
+- 3D: Hybrid SDF/raster, BVH acceleration, physics system
+- Shaders: Modular WGSL, comprehensive compilation tests
+```
+
+### Bad: Verbose PROJECT_CONTEXT.md Status
+```markdown
+### Recently Completed
+
+#### Audio Peak Measurement (February 7, 2026)
+- **Real-Time Peak Fix**: Fixed critical bug where... [20 lines]
+- **Peak Decay Optimization**: Changed decay from 0.95 to 0.7... [15 lines]
+- **SilentBackend**: Created test backend with... [25 lines]
+- [200+ more lines of detailed milestones]
+```
+
+---
+
+## Questions?
+
+If unsure whether to archive something, ask:
+
+1. **Is this about work completed >30 days ago?** → Archive
+2. **Is this a detailed implementation plan?** → Archive (keep 2-line summary)
+3. **Is this needed for next week's work?** → Keep
+4. **Is this >50 lines of historical detail?** → Archive
+5. **Would a new contributor need this immediately?** → Keep if yes
+
+When in doubt, **archive with a pointer**. Better to have clean context and occasionally load an archive than to bloat every session.
+
+---
+
+## Maintenance Checklist
+
+Copy this for monthly reviews:
+
+```markdown
+## Context Maintenance - [Month YYYY]
+
+### File Sizes (before)
+- [ ] PROJECT_CONTEXT.md: ___ lines
+- [ ] TODO.md: ___ lines
+- [ ] Tier 1-2 total: ___ words
+
+### Actions Taken
+- [ ] Archived tasks >30 days old to COMPLETED.md
+- [ ] Moved temp analysis docs to doc/
+- [ ] Updated PROJECT_CONTEXT.md current status
+- [ ] Simplified TODO.md task descriptions
+- [ ] Verified CLAUDE.md/GEMINI.md config
+- [ ] Updated cross-references
+
+### File Sizes (after)
+- [ ] PROJECT_CONTEXT.md: ___ lines (target: <300)
+- [ ] TODO.md: ___ lines (target: <500)
+- [ ] Tier 1-2 total: ___ words (target: <10K)
+
+### Notes
+[Any issues encountered, files moved, etc.]
+```
+
+---
+
+*Last updated: February 7, 2026*
diff --git a/GPU_EFFECTS_TEST_ANALYSIS.md b/doc/GPU_EFFECTS_TEST_ANALYSIS.md
index f5b74ce..f5b74ce 100644
--- a/GPU_EFFECTS_TEST_ANALYSIS.md
+++ b/doc/GPU_EFFECTS_TEST_ANALYSIS.md
diff --git a/HANDOFF.md b/doc/HANDOFF.md
index bc22314..bc22314 100644
--- a/HANDOFF.md
+++ b/doc/HANDOFF.md
diff --git a/HANDOFF_2026-02-04.md b/doc/HANDOFF_2026-02-04.md
index f38e5be..f38e5be 100644
--- a/HANDOFF_2026-02-04.md
+++ b/doc/HANDOFF_2026-02-04.md
diff --git a/doc/HANDOFF_CLAUDE.md b/doc/HANDOFF_CLAUDE.md
new file mode 100644
index 0000000..24de0b1
--- /dev/null
+++ b/doc/HANDOFF_CLAUDE.md
@@ -0,0 +1,286 @@
+# Handoff Summary - Claude to Gemini
+
+**Date:** February 7, 2026
+**Session:** Audio Peak Measurement & Test Coverage Improvements
+**Branch:** main (1 commit ahead of origin)
+**Commit:** a6a7bf0
+
+---
+
+## Work Completed ✅
+
+### Critical Audio-Visual Sync Fix
+**Problem:** Visual effects (screen flashes) triggered ~400ms before audio was actually heard, causing poor synchronization.
+
+**Root Cause:** Peak measurement occurred at ring buffer write time (synth_render) rather than playback time (audio callback).
+
+**Solution:**
+- Added `get_realtime_peak()` to AudioBackend interface
+- Implemented real-time peak tracking in MiniaudioBackend audio_callback
+- Used exponential averaging: instant attack, 0.7 decay rate (1-second fade)
+- Updated main.cc and test_demo.cc to use `audio_get_realtime_peak()`
+
+**Impact:** Visual effects now sync perfectly with heard audio.
+
+---
+
+### Peak Decay Rate Optimization
+**Problem:** Initial decay rate (0.95) caused 5.76-second fade times, making test_demo "just flash constantly" (user's words).
+
+**Analysis:**
+- Old: 0.95^45 ≈ 0.1 at 45 callbacks (128ms each) = 5.76 seconds
+- New: 0.7^9 ≈ 0.1 at 9 callbacks = 1.15 seconds
+
+**Impact:** Screen flashes now have proper attack/decay for responsive visual sync.
+
+---
+
+### SilentBackend for Testing
+**Created:** `src/audio/backend/silent_backend.{h,cc}` + 7 comprehensive tests
+
+**Features:**
+- Test-only backend with no audio output
+- Controllable peak for edge case testing
+- Tracks frames rendered and voice triggers
+- Integration with audio.cc for coverage improvement
+
+**Tests Added:**
+1. Backend lifecycle (init/start/shutdown)
+2. Audio system integration
+3. Peak control
+4. Frame/voice tracking
+5. Playback time advancement
+6. Buffer management
+7. audio_update() safety
+
+**Coverage:** Significantly improved audio.cc test coverage (audio_get_playback_time, audio_render_ahead, audio_update all tested).
+
+---
+
+### Backend Reorganization
+**Change:** Moved all backend implementations to `src/audio/backend/` subdirectory
+
+**Files Moved:**
+- miniaudio_backend.{h,cc}
+- mock_audio_backend.{h,cc}
+- wav_dump_backend.{h,cc}
+- jittered_audio_backend.{h,cc}
+- silent_backend.{h,cc}
+
+**Kept:** audio_backend.h (interface) remains in src/audio/
+
+**Updates:**
+- All #include paths updated in backend and test files
+- CMakeLists.txt AUDIO_SOURCES and test targets updated
+- Relative includes ("../audio.h", "../ring_buffer.h") in backend .cc files
+
+---
+
+### Dead Code Removal
+**Removed:** `register_spec_asset()` function from audio.{h,cc}
+- Function was never called anywhere in codebase
+- Declaration + implementation (lines 43-58) removed
+
+---
+
+## Test Status
+
+**All Tests Passing:** 28/28 (100%)
+
+**Test Breakdown:**
+- HammingWindowTest ✓
+- MathUtilsTest ✓
+- AssetTest ✓
+- ProceduralTest ✓
+- DctTest ✓
+- FftTest ✓
+- AudioGenTest ✓
+- SynthEngineTest ✓
+- AudioBackendTest ✓
+- MockAudioBackendTest ✓
+- WavDumpTest ✓
+- JitteredAudioTest ✓
+- SilentBackendTest ✓ (NEW)
+- TrackerTest ✓
+- TrackerTimingTest ✓
+- VariableTempoTest ✓
+- SequenceSystemTest ✓
+- DemoEffectsTest ✓
+- test_3d_render ✓
+- demo64k ✓
+
+**No Regressions:** All existing tests continue to pass.
+
+---
+
+## Build Status
+
+**Configuration:**
+```bash
+cmake -S . -B build -DDEMO_ALL_OPTIONS=ON
+cmake --build build
+cd build && ctest
+```
+
+**Results:**
+- Build: ✓ Clean (no errors, no warnings)
+- Tests: ✓ 28/28 passing
+- Executables: ✓ demo64k and test_demo run successfully
+
+**Known Pre-Existing Issues (Not Fixed):**
+- test_effect_base: Missing WebGPUTestFixture symbols (unrelated)
+- test_post_process_helper: Missing main() (unrelated)
+
+---
+
+## Files Modified
+
+**New Files (3):**
+- src/audio/backend/silent_backend.h
+- src/audio/backend/silent_backend.cc
+- src/tests/test_silent_backend.cc
+
+**Modified Files (15):**
+- src/audio/audio_backend.h (added get_realtime_peak)
+- src/audio/backend/miniaudio_backend.{h,cc} (peak tracking)
+- src/audio/backend/mock_audio_backend.{h,cc} (interface update)
+- src/audio/backend/wav_dump_backend.{h,cc} (interface update)
+- src/audio/backend/jittered_audio_backend.{h,cc} (interface update)
+- src/audio/audio.{h,cc} (added audio_get_realtime_peak, removed register_spec_asset)
+- src/audio/synth.h (added deprecation note)
+- src/main.cc (use audio_get_realtime_peak)
+- src/test_demo.cc (use audio_get_realtime_peak)
+- src/tests/test_audio_backend.cc (TestBackend update)
+- CMakeLists.txt (paths and new test)
+
+**Moved Files (8):**
+- src/audio/miniaudio_backend.{h,cc} → src/audio/backend/
+- src/audio/mock_audio_backend.{h,cc} → src/audio/backend/
+- src/audio/wav_dump_backend.{h,cc} → src/audio/backend/
+- src/audio/jittered_audio_backend.{h,cc} → src/audio/backend/
+
+**Documentation (2):**
+- PEAK_FIX_SUMMARY.md (technical analysis)
+- TASKS_SUMMARY.md (task completion report)
+
+---
+
+## Git Status
+
+**Branch:** main
+**Status:** Working tree clean
+**Unpushed Commits:** 1 commit (a6a7bf0)
+
+**Last Commit:**
+```
+commit a6a7bf0
+Author: skal <skal@example.com>
+Date:   Thu Feb 7 2026
+
+    feat(audio): Add SilentBackend, fix peak measurement, reorganize backends
+
+    This commit completes 5 tasks:
+    1. Fix real-time audio peak measurement for visual sync
+    2. Create SilentBackend for audio testing
+    3. Reorganize audio backends to audio/backend/ directory
+    4. Remove dead code (register_spec_asset)
+    5. Add comprehensive tests for audio.cc coverage
+
+    [Full commit message in git log]
+```
+
+**Next Step:** `git push` to sync with remote
+
+---
+
+## Technical Details
+
+### AudioBackend Interface Change
+```cpp
+// New pure virtual method
+virtual float get_realtime_peak() = 0;
+```
+
+**Implementations:**
+- MiniaudioBackend: Measures peak in audio_callback (exponential averaging)
+- MockAudioBackend: Returns 0.5f (test fixture)
+- WavDumpBackend: Returns 0.0f (no playback)
+- JitteredAudioBackend: Delegates to inner backend
+- SilentBackend: Returns controllable test_peak_
+
+### Peak Measurement Algorithm
+```cpp
+// In MiniaudioBackend::audio_callback (after ring_buffer->read)
+float frame_peak = 0.0f;
+for (int i = 0; i < actually_read; ++i) {
+  frame_peak = fmaxf(frame_peak, fabsf(fOutput[i]));
+}
+
+// Exponential averaging
+if (frame_peak > realtime_peak_) {
+  realtime_peak_ = frame_peak;  // Attack: instant
+} else {
+  realtime_peak_ *= 0.7f;       // Decay: 30% per callback (~1 second fade)
+}
+```
+
+### SilentBackend Key Methods
+```cpp
+void init() override;           // Sets initialized_ = true
+void start() override;          // Sets started_ = true
+void shutdown() override;       // Resets flags
+float get_realtime_peak() override;  // Returns test_peak_
+void on_frames_rendered(int num_frames) override;  // Tracks rendering
+void on_voice_triggered(...) override;  // Tracks triggers
+
+// Test inspection interface
+void set_peak(float peak);
+void reset_stats();
+int get_frames_rendered() const;
+int get_voice_trigger_count() const;
+```
+
+---
+
+## Context for Next Session
+
+### Current State
+- Audio system is fully functional with proper visual sync
+- Test infrastructure significantly improved
+- Code organization cleaner (backend subdirectory)
+- All tests passing, no known regressions
+
+### Potential Next Steps
+1. **Task #5: Spectral Brush Editor** - Top priority, in progress
+2. **Task #18: 3D System Enhancements** - OBJ asset pipeline
+3. **Continue test coverage improvements** - Other subsystems
+4. **Size optimization** - Once feature-complete
+
+### Notes for Gemini
+- Peak measurement fix was critical user-reported bug ("test_demo just flashing")
+- SilentBackend pattern can be reused for other subsystems needing test infrastructure
+- Backend organization improves maintainability but doesn't affect binary size
+- All audio functionality is tested and working correctly
+
+---
+
+## User Feedback Summary
+
+**Session Start:** User identified missing audio.cc coverage and timing bug
+**Critical Bug Report:** "test_demo is just flashing" - peak decay too slow
+**Outcome:** User satisfied, requested commit (completed)
+
+---
+
+## Handoff Checklist
+
+- [x] All tests passing (28/28)
+- [x] Working tree clean
+- [x] Documentation updated (PEAK_FIX_SUMMARY.md, TASKS_SUMMARY.md)
+- [x] Commit created with detailed message
+- [x] No known regressions
+- [x] Ready for next session
+
+---
+
+**handoff(Claude):** Audio peak measurement fixed, test coverage improved, backend reorganized. All 28 tests passing. Ready for push and next task.
diff --git a/PEAK_FIX_SUMMARY.md b/doc/PEAK_FIX_SUMMARY.md
index cf42233..cf42233 100644
--- a/PEAK_FIX_SUMMARY.md
+++ b/doc/PEAK_FIX_SUMMARY.md
diff --git a/PLATFORM_ANALYSIS.md b/doc/PLATFORM_ANALYSIS.md
index eefbd42..eefbd42 100644
--- a/PLATFORM_ANALYSIS.md
+++ b/doc/PLATFORM_ANALYSIS.md
diff --git a/PLATFORM_SIDE_QUEST_SUMMARY.md b/doc/PLATFORM_SIDE_QUEST_SUMMARY.md
index d4be581..d4be581 100644
--- a/PLATFORM_SIDE_QUEST_SUMMARY.md
+++ b/doc/PLATFORM_SIDE_QUEST_SUMMARY.md
diff --git a/TASKS_SUMMARY.md b/doc/TASKS_SUMMARY.md
index 77cde6b..77cde6b 100644
--- a/TASKS_SUMMARY.md
+++ b/doc/TASKS_SUMMARY.md
diff --git a/test_demo_README.md b/doc/test_demo_README.md
index bbce023..bbce023 100644
--- a/test_demo_README.md
+++ b/doc/test_demo_README.md