feat: provenance de carpeta vs Workspace en 3D y Director

[IAnMove]

Objetivo

Aplica el contrato de dominio de #86 al flujo 3D + Director. PR cohesivo de writers de servicio (no _launch_runtime.py). +691/−23.

Contrato

• workspace_id = ID opcional de colección Workspace. No se inventa a partir del nombre de un directorio.
• output_folder = nombre de la carpeta de salida física (nunca una ruta absoluta del host).
• El campo workspace de jobs/pipelines en este código es el nombre de carpeta → se pasa como output_folder=.
• Iniciador = origin.actor / tool / capability. Actor ausente permanece unknown, nunca se inventa user.
• Proveedor/modelo = generation.model, no origin.
• Tras el mesh 3D/rig en disco, un fallo de sidecar no borra el GLB (publish_generation_sidecar_best_effort).
• CommandContext: job_id, task_id, root_task_id, pipeline_id en correlaciones del sidecar → execution.*. El pipeline id de Director sigue forzando tool=director.

Writers tocados

• app/services/model3d_service.py — Hunyuan3D: output_folder= + model_type/provider en params
• app/services/rig_service.py — mismo patrón, tool=rig
• app/services/director_pipeline.py — _write_director_assembly_sidecar y persist_pipeline_output_timing
• app/services/alternative_songs.py — save_sidecar / remount: carpeta vs colección real (workspace_…); copia IDs de pipeline Director

No se toca Studio generate, Wizard, Series Lab, Comics ni _launch_runtime.py.

Tests

• tests/test_model3d_rig_task_identity.py — carpeta sin workspace_id, IDs en execution, GLB sobrevive si el publish interno lanza
• tests/test_director_pipeline_timing.py — assembly/timing: output_folder, pipeline_id/job_id/task_id
• tests/test_alternative_songs.py — folder-only, colección real distinta, IDs de Director en remount
• tests/test_provenance_3d_director.py — writers vía helpers / sidecar parcheado
• tests/test_asset_manifest.py y scripts/architecture_contracts.py — sin cambios de contrato

Local: 64 passed. python scripts/architecture_contracts.py ok.

No mezclar hasta CI verde.

---

Note

Medium Risk
Changes gallery manifest origin/execution fields across multiple production paths; incorrect prior workspace_id usage could affect downstream provenance consumers, but scope is metadata writers with broad test coverage and no auth or payment logic.

Overview
Aligns 3D, Rig, Director, and alternative-songs sidecar writers with the asset-manifest provenance contract: workspace_id is only an optional Workspace collection (workspace_…), and output_folder is the physical output directory basename (never a host path).

Hunyuan3D and rig jobs now publish with output_folder= derived from the job’s workspace string, plus model_type / provider on generation params. Director assembly and timing persistence do the same, and they record pipeline_id, job_id, task_id, and root_task_id on manifests for execution.* correlation.

Alternative songs resolves folder vs collection via _sidecar_location (folder names are not invented as collections; real collection IDs are preserved on re-save). Remount sidecars copy Director pipeline/task IDs and default remount provider to ffmpeg.

Tests are updated and test_provenance_3d_director.py adds end-to-end checks (absolute paths stripped, folder-only origin, provenance round-trip).

^{Reviewed by Cursor Bugbot for commit f7aad1e. Configure here.}

[github-actions]

PR Review — Loreframe Studio

Risk: medium
Scope: 8 file(s); +691/-23; backend services

Automated review from scripts/analyze_pr.py. This is a heuristic pass (no LLM) so humans still own the merge decision.

Findings

• medium — Very large file change (tests/test_provenance_3d_director.py)
  tests/test_provenance_3d_director.py adds 406 lines. Consider splitting the PR.

Changed files

• added: tests/test_provenance_3d_director.py
• modified: app/services/alternative_songs.py, app/services/director_pipeline.py, app/services/model3d_service.py, app/services/rig_service.py, tests/test_alternative_songs.py, tests/test_director_pipeline_timing.py, tests/test_model3d_rig_task_identity.py

CONTRIBUTING checklist

• python scripts/verify_clean_repo.py
• python -m compileall -q app/services app/launch.py scripts
• cd ui && npm run build if the UI changed
• No weights, CivitAI sidecars, or generated guides
• Stays local-first (no required accounts / telemetry)

Posted by the repo PR review workflow. Re-runs on each push to the PR.

[IAnMove]

@cursor review

[cursor]

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit f7aad1e. Configure here.}