This workspace has been restructured around a smaller active surface:
core/contains the sequencing engine, audio analysis, model parsing, and build orchestration.xlights/contains the legacy-derived XSQ writer helpers and the effect catalog cache.tools/contains shared utilities and preview rendering.ai/contains explicit opt-in bridge stubs for future model integrations.archive/legacy_versions/preserves older wrappers, lab scripts, and legacy entrypoints.
core/
ai/
xlights/
tools/
tests/
archive/legacy_versions/
main.py
README.md
requirements.txt
AGENTS.md
TASKS.md
ROADMAP_BETA_TODO.md
The current beta-readiness plan is tracked in:
ROADMAP_BETA_TODO.md— autonomous beta-version roadmap and task order.TASKS.md— short entrypoint for humans and coding agents.docs/SUPPORT_MATRIX.md— supported beta platforms, inputs, and boundaries.docs/BETA_POLICY.md— data-use, learning, and asset-safety policy.
Beta expectations:
- Use copies of layouts, templates, and audio inputs.
- Treat generated output as beta-quality until manually reviewed in xLights.
- Do not commit private tester files, songs, layouts, templates, screenshots, or generated outputs unless explicitly approved.
- The near-term goal is a safe, inspectable beta GUI path, not perfect production-quality auto-sequencing.
Launch the maintained GUI control center (preferred day-to-day workflow):
python gui_launcher.pyor use:
launch_sequencer_app.cmdList active sequencing profiles:
python main.py --list-profilesRun the active master profile and pass through engine arguments:
python main.py --profile master -- --template template.xsq --audio 13.wav --no-promptLegacy version IDs still work as explicit compatibility fallbacks:
python main.py --profile v27.3 -- --template template.xsq --audio 13.wav --no-promptBuild or refresh the dedicated Helixville 3D test show folder from your allmodels / GP baseline:
python tools/build_helixville_layout.pyOutput is written to helixville/, including:
helixville/xlights_rgbeffects.xmlhelixville/xlights_rgbeffects.xbkp(when source backup exists)helixville/helixville_manifest.json
Build the aligned legacy-256 based Helixville3 layout with lyric/singing-face groups:
python tools/build_helixville3_layout.pyOutput is written to helixville3/, including:
helixville3/xlights_rgbeffects.xmlhelixville3/helixville3_manifest.jsonhelixville3/HELIXVILLE3_LAYOUT_NOTES.txt
Legal source notes:
docs/helixville3_legal_sources_2026-04-23.md
Enable the AC-first hardKor placement machine:
python main.py --profile v27.3 -- --template template.xsq --audio 13.wav --hardkor --ac-lights-only --no-prompthardKor rules and source references:
docs/hardkor_rulebook_2026-04-23.md
Generate a multi-variant comparison pack (.xsq + .mp4) using 13.wav:
python tools/generate_aaatest_pack.pyOutputs are written to aaatest/.
- The active maintained entrypoint is the
masterprofile, currently backed by the stablev27.3tuning insidecore/effect_engine.py. - Legacy wrappers and experimental scripts were moved instead of deleted.
- The current restructure is intentionally conservative: proven engine code was promoted into the new folders with minimal behavioral changes.
- AI bridges are placeholders by design. Rule-based sequencing remains the default path.
Helix scoring memory is limited to Helix-generated sequence reports that carry the internal Dream Sequence Weaver watermark metadata. It does not learn from copyrighted songs, licensed third-party sequences, template files, imported third-party XSQs, or arbitrary user-provided sequence folders. Audio analysis may guide a one-time render, but persistent learning stores only Helix decisions, generated-score metrics, and generated sequence context.