I also have an opengraph image to add to the head, but I'd rather wait for #9161 to get merged to avoid merge conflicts

Hold off on merging, found some docs additions that someone put in the wrong places.

• invokeai/docs/src/config/redirects.ts - the node redirects need correction.

  • /nodes/NODES, /nodes/comfyToInvoke, and /nodes/communityNodes redirect to /workflows/..., but the migrated pages are under /features/workflows/....
  • These old links will 404 unless the redirect targets are updated.
  • Add a redirect-target validation test so every configured redirect resolves to an existing generated docs route.

• invokeai/docs/src/content/docs/concepts/models.mdx - the orphaned model directory sync/delete workflow still needs documentation.

  • The code still exposes this feature through OrphanedModelsService, /api/v2/models/sync/orphaned, SyncModelsButton, and SyncModelsDialog.
  • The current docs only cover "Missing Files", which means database entries whose files are gone.
  • That is different from the sync workflow, which finds model directories on disk that are not referenced in the database.
  • The restored docs should be updated for current behavior: .gguf is included, the UI is admin-only, and deletion directly removes directories with shutil.rmtree().

• invokeai/docs/src/content/docs/features/Canvas/canvas-projects.mdx - canvas projects still need developer documentation, not just user-facing docs.

  • The code confirms the technical format is current: .invk ZIP files contain manifest.json, canvas_state.json, params.json, ref_images.json, loras.json, and images/.
  • The implementation still has manifest version 1, image-name collection, server-existence checks, upload/remapping for missing images, and concurrency-limited image fetch/upload.
  • The old technical docs should be migrated under development docs, but updated to use the current UI entry points: the canvas toolbar archive menu and canvas context menu.

• invokeai/docs-old/nodes/detailedNodes/faceTools.md - face Nodes still need a replacement docs page.

  • The code still defines FaceOff, FaceMask, and FaceIdentifier in facetools.py, category segmentation, version 1.2.2.
  • The old page is mostly still accurate: face IDs, minimum confidence, offsets, padding, chunking, invert mask, and outputs match the current code.
  • Refresh the wording against current field descriptions, then add a redirect from /nodes/detailedNodes/faceTools.

@lstein this is all good to go