Merge branch 'ui-stabilization/phase-0' into develop

UI stabilization Phase 0: ratchet and baseline, .svelte counted in the
gate report, U1-U5 fixed, twin pins, seam-map rows.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

Author: dkackman

.github/workflows/ci.yml

@@ -69,9 +69,11 @@ jobs:
           cache-dependency-path: ui/package-lock.json
       - run: npm ci
       - name: Format check
-        run: npx prettier --check src e2e *.ts *.js
+        run: npx prettier --check src e2e scripts *.ts *.js
       - name: Lint
         run: npm run lint
+      - name: Architecture ratchet
+        run: npm run metrics -- --check ../docs/stabilization/ui/baseline.json
       - name: Type check
         run: npm run check
       - name: Unit tests

docs/ARCHITECTURE.md

@@ -150,3 +150,7 @@ one to open.
 | Concept | Owner | Rule | Enforced by |
 | --- | --- | --- | --- |
 | The UI reads engine fields | `ui/src/lib/plan.ts`: `describePlan`, `ui/src/lib/results.ts`: `sectionBySubfolder` | The UI reads `plan`, `version` and `subfolder` as fields the server sends and derives nothing of its own, and it never sends `acknowledged_cost`. | `ui/src/lib/plan.test.ts`, `ui/src/lib/results.test.ts` |
+| Reference prefixes in the UI | `ui/src/lib/references.ts` | The only module in `ui/src` that spells a reference prefix; each equals `dw/references.py`'s. | `tests/test_ui_twins.py::test_the_ui_spells_every_reference_prefix_the_engine_does`; `prefix_literals` in `ui/scripts/arch-metrics.mjs` |
+| Output kinds on the job page | `dw/server/outputs.py`: `MEDIA_KINDS`, `output_kinds` | The page renders an output by the `output_kinds` the job detail carries, never by its extension. | `tests/test_server.py::test_a_job_reports_each_output_files_media_kind`, `ui/src/lib/pages/JobPage.test.ts` |
+| UI copies of engine rules | the constants `tests/test_ui_twins.py` reads | A list the UI must hold equals its owner, read from the TS source; a rule tested on both sides reads one case file in `tests/fixtures/`. | `tests/test_ui_twins.py` |
+| UI architecture ratchet | `ui/scripts/arch-metrics.mjs` | No metric rises above `docs/stabilization/ui/baseline.json`. | the `ui` job in `.github/workflows/ci.yml`; `npm run preflight` |

docs/stabilization/ROADMAP.md

@@ -827,7 +827,9 @@ The gates measured `dw_mcp` and `ui/src` but never read them. Two follow-up
 surveys do: [mcp-assessment.md](mcp-assessment.md) and
 [ui/ASSESSMENT.md](ui/ASSESSMENT.md). Every gate's UI SLOC row above counts only `.ts`:
 pygount has no lexer for `.svelte` and reports those files as 0 lines, so
-`ui/src` is about 15,500 raw lines, not the ~2,270 code lines shown.
+`ui/src` is about 15,500 raw lines, not the ~2,270 code lines shown. With `.svelte` counted (`scripts/arch_report.py`'s
+`svelte_code_lines`), the UI is 12,969 code lines at `stabilization-gate-4`,
+against 20,362 for the engine and 6,956 for the API.
 
 ## Working rules for the duration
 

docs/stabilization/ui/ROADMAP.md

@@ -21,15 +21,13 @@ works on is what the earlier phases leave behind. Phase 0's plan is
 
 ## Decisions
 
-Recorded 2026-10-01. The two marked **open** are Don's.
+Recorded 2026-10-01; Don ruled on both open questions the same day.
 
-- **Freeze (open).** The engine pass froze features until its guardrails
+- **Freeze (yes, Don 2026-10-01).** The engine pass froze features until its guardrails
   existed. The UI equivalent: no new pages, components or UI features until
   gate 2; the implementer fixes tester bugs in `ui/` through existing code,
   and a fix that needs a new component waits or is labelled `stabilization`.
-  Proposed rather than assumed, because the harness's tester bugs land in the
-  UI less often than in the engine.
-- **Component library (open; recommended: Bits UI).** Headless, so the
+- **Component library: Bits UI (approved, Don 2026-10-01).** Headless, so the
   `app.css` token system, the achromatic chrome and the WCAG AA pairs in
   `ui/CLAUDE.md` stay as they are. It is native to Svelte 5 runes, covers every
   hand-built widget (dialog, alert dialog, popover, combobox, tooltip, toggle

docs/stabilization/ui/baseline.json

@@ -0,0 +1,9 @@
+{
+  "files_over_size_ceiling": 7,
+  "complex_functions": 14,
+  "long_functions": 2,
+  "import_cycles": 1,
+  "modules_in_import_cycles": 2,
+  "prefix_literals": 15,
+  "a11y_suppressions": 7
+}

docs/stabilization/ui/phase-0.md

@@ -12,7 +12,7 @@
 
 ## Global Constraints
 
-- Freeze, if Don confirms it (ROADMAP *Decisions*): no new UI features. Change only what a task names.
+- Freeze (ROADMAP *Decisions*, Don 2026-10-01): no new UI features. Change only what a task names.
 - The only new modules are `ui/scripts/arch-metrics.mjs`, `ui/src/lib/references.ts`, test files and `tests/fixtures/*.json`.
 - Every new test fails on the code before its fix. Run it and see the failure before writing the fix.
 - No count-pinning assertions. Baseline numbers live in `docs/stabilization/ui/baseline.json`, never in a test.

dw/server/jobs.py

@@ -59,6 +59,7 @@
     TERMINAL_STATES,
     Job,
 )
+from .outputs import output_kinds
 
 logger = logging.getLogger("dw")
 
@@ -702,6 +703,9 @@ def _record_progress(self, job, event):
             event["files"] = self._relative_output_names(
                 event["files"], job.spec.get("output_dir")
             )
+            # A running job's page renders each output as its step ends,
+            # before there is a manifest to classify
+            event["output_kinds"] = output_kinds([{"files": event["files"]}])
         if "manifest" in event:
             # workflow_end carries the run's full manifest nested under this
             # key - it must match get_job's rendering of the same list rather

dw/server/outputs.py

@@ -56,6 +56,20 @@
     ".txt": "text",
 }
 
+
+def output_kinds(manifest):
+    """Each file a job manifest lists, mapped to its MEDIA_KINDS entry, or
+    None for a kind the gallery does not show. A client renders an output
+    by this rather than keeping its own extension list."""
+    kinds = {}
+    for entry in manifest or []:
+        if not isinstance(entry, dict):
+            continue
+        for name in entry.get("files") or []:
+            kinds[name] = MEDIA_KINDS.get(os.path.splitext(name)[1].lower())
+    return kinds
+
+
 # The allowlist members that are not already-compressed containers -
 # everything else in MEDIA_KINDS deflates for about nothing, so it is
 # stored instead (see zip_download)

dw/server/routes/jobs.py

@@ -55,7 +55,12 @@
     RUNNING,
     TERMINAL_STATES,
 )
-from ..outputs import absolute_served_url, asset_library_for_job, served_url
+from ..outputs import (
+    absolute_served_url,
+    asset_library_for_job,
+    output_kinds,
+    served_url,
+)
 
 logger = logging.getLogger("dw")
 
@@ -226,7 +231,8 @@ def get_job(request: Request, job_id: str):
     if job is None:
         raise HTTPException(status_code=404, detail="Unknown job")
     # a historical job is already a detail dict; a live one renders itself
-    return job if isinstance(job, dict) else manager.describe(job)
+    detail = job if isinstance(job, dict) else manager.describe(job)
+    return {**detail, "output_kinds": output_kinds(detail.get("manifest"))}
 
 
 @router.get("/api/jobs/{job_id}/workflow")

scripts/arch_report.py

@@ -15,6 +15,7 @@
 import collections
 import itertools
 import pathlib
+import re
 import statistics
 import subprocess
 import sys
@@ -98,6 +99,23 @@ def extract(ref, root, into):
             tar.extractall(into, filter="data")
 
 
+# A line that is only a comment. A multi-line /* */ comment's inner lines
+# count as code: matching them by a leading '*' would also match a CSS '*'
+# selector, and an undercount hides growth where an overcount does not.
+SVELTE_COMMENT = re.compile(r"^(<!--.*-->|//.*|/\*.*\*/)$")
+
+
+def svelte_code_lines(path):
+    """Code lines in a .svelte file, which pygount reads as 0: every line
+    that is neither blank nor a comment on its own. Script, markup and style
+    all count, since all three are what a reader holds."""
+    return sum(
+        1
+        for line in path.read_text(encoding="utf-8").splitlines()
+        if line.strip() and not SVELTE_COMMENT.match(line.strip())
+    )
+
+
 def sloc(tree):
     from pygount import SourceAnalysis
 
@@ -110,7 +128,12 @@ def sloc(tree):
                 and keep(path.relative_to(tree))
                 and not any(part in EXCLUDED for part in path.parts)
             ):
-                counts[layer] += SourceAnalysis.from_file(str(path), layer).code_count
+                if path.suffix == ".svelte":
+                    counts[layer] += svelte_code_lines(path)
+                else:
+                    counts[layer] += SourceAnalysis.from_file(
+                        str(path), layer
+                    ).code_count
     return counts