Skip to content

Build order — which milestone the harness works, top to bottom

ONE ordered list, read by builder milestone, builder queue and the loop (spec harness/issues-and-labels.md): the harness briefs the first milestone here that still has open issues, unless a person names one. The owner orders the whole list, by dragging it in the console (builder status, Milestones), which commits this file on branch build-order as a tiny batch the loop checks and lands; the order the loop uses is the landed one. builder triage --order proposes additions at the end, as a diff, never writes it. Parked is not ordered: buckets and lists that are not milestones.

In order

  1. Builder — the harness itself (the milestone was dev-workflows in this file and dev-orchestration-harness on GitHub until the Phase 0 batch); runs beside the product lane. Spec: harness/building-fichero-locally.md
  2. image-enhancement — the one product lane tonight (Step 1.1 landed first). Spec: source/image-enhancement.md
  3. Sharing & Pairing — Dead-Simple UX — spec APPROVED with 27 steps; rename the milestone to sharing-two-macs when triage runs. Spec: source/sharing-two-macs.md
  4. recipe-designer — wide onboarding; the recipe phases R1–R5 go into its build order first. Spec: source/wide-onboarding.md
  5. recipes-run — running things and watching them: a recipe run from Start to the summary. Spec: source/recipes-run.md; how any run fails, pauses and goes on: compute/runner-reliability.md
  6. activity — running things and watching them: the Activity pane and window, the one job model. Spec: ui/activity-pane.md
  7. workflows — running things and watching them: the workflow editor and bar. Spec: ui/workflows.md
  8. workflow-node-config — running things and watching them. Spec: ui/workflow-node-config.md
  9. source-model — the data model and getting material in
  10. importer — the data model and getting material in
  11. undo-trash-rollback — the data model and getting material in
  12. panes-workspaces — the shell
  13. modes-to-panes — the shell
  14. library-view-modes — the shell
  15. sidebar-crud — the shell
  16. menus-and-commands — the shell
  17. preview-surface — the shell
  18. reader-view — the shell
  19. kg-tables — knowledge and finding
  20. kg-readable-representation — knowledge and finding
  21. kg-entity-inspector — knowledge and finding
  22. kg-enrichment — knowledge and finding
  23. search — knowledge and finding
  24. explore — knowledge and finding
  25. research — knowledge and finding
  26. ai-settings — engine and providers
  27. remote-compute — engine and providers
  28. transport-http-uds — engine and providers
  29. engine-startup-lifecycle — engine and providers
  30. audited-action-layer — engine and providers
  31. observable-data-layer — engine and providers
  32. export-pipeline — output
  33. historical-text-normalization — output
  34. test-environment-contract — testing and release
  35. ui-testing-strategy — testing and release
  36. ui-test-harness — testing and release
  37. release-and-versioning — testing and release
  38. cloud-permission-and-spend — caps in Settings, images to the cloud, a refusal with a way through. Spec: ai/cloud-permission-and-spend.md

Parked

  • Next release (98 issues) and cand 16 (18) — grab-bags, not specs: triage files each issue into its spec’s milestone and these close.
  • Bugs - *, Performance - *, Lint - Structural Debt - * — ops buckets, worked by bug and priority, never as a lane.
  • UX - *, Engine - *, App - *, Developer - *, Hygiene & Structure - *, SwiftUI/Engine - *, Mac App Store — *, Testing Overhaul — *, Inspector View, Library View, Keyboard Navigation, Search View - *, Sidebar & Launch Performance — older area buckets, not specs: worked by bug and priority (lead ruling 2026-11-02).
  • cand 14, Later (post-release), iPhone, HF, ACENet — not worked now (lead ruling 2026-11-02).