T098: display harnesses (A), and the server's scope labels from the served facet (plan 038, T094's R2-INV-R9)

[brettheap]

Lane: openxfactory-4 (openXfactory-4-openDox_extraction)
Authored by lane openXfactory-3, delegated by lane openxfactory-4 (#656 6021862323)

Arc: neutral-product-standalone-operability

T098: display harnesses (A), and the server's scope labels from the served facet (plan 038)

This is a slice of T094's map R2-INV-R9 (opensoft/openxFactory#656 6021830607). tasks.md has no entry for it yet: lane 4 records the T095 to T104 entries after T002 (#1251). Until then, its contract is the map's slice-outline row plus the holder's rulings, 6021830531.

• Realizes: 21 of the composed reds outside 12.5's set in the declared exclusion, toward requirement 9's declared composed tests (T073, F9.1):

  • canvas: DJ 8 and DJ + HR 1;
  • explorer: DJ 5 and CP 2;
  • the 5 nodes held by R9-R1: canvas 2, explorer 2, scope 1.

• Falsifier (the outline row): the composed run of the three files, each alone, at main + T095, takes the named 21 to 0 red. These read unchanged:

  • 12.5's 16 governed suites;
  • F5.2's 5 suites;
  • test_column_contributions_governed.py (9 passed).

  tests/test_declared_exclusion.py reads green when run lone.

• Ruled:

  • R9-R1 (a): doxbench_scope.py's labels and notes derive from the facet the browser reads, and every pin that asserts words is respelled through VR to the SERVED words.
  • R9-W1 (A″): the harnesses copy display.js beside the module they copy, as T026, U-7: the allow-list for 12.5's governed suites, with R-1 (a)'s span kind (plan 038, H-2) #43 does.
  • Both are in 6021830531. They conform to the ruled overlay (5784683830, 5801057769) and to the served-display rule for the staging suite (6016648451).

• After: T095 (T095: the composed harness for the declared integration tests (plan 038, T094's R2-INV-R9) #46), LANDED as e2340bb6. The served display needs the governed host that T095's DECLARED_HOST_SUITES registers for these three files when the run is composed. This branch merges main e2340bb6, and every number below is measured over it.

What changes

src/openxdox/doxbench_scope.py is this slice's under R9-R1 (a), as claim 6021862323 names it.

• The labels now come from the facet. The section labels and notes were one domain's words ("cluster documents", "topic folder documents"). They are now built from the facet by _section_meta(_served_display()) once per resolve_scope call.
  • The six sentences are word for word the ones openDox's staging model has built since § 3.4 slice S7 (views/staging-workbench-model.js SECTION_META).
  • Keys, flags and paths are unchanged.
• Where the words come from. _served_display() is the display block that /capabilities serves: display_manifest(host_display()).
  • It is read at call time, never at import and never cached, as opendox.canvas_drafts reads its one shared sentence.
• Decision: no host means neutral words. With no host registered there is no served page. The browser model given no facet speaks openDox's neutral words, so the server answers with those too rather than refusing.
  • The words label sections of an authorization answer and never decide it.
  • A library or test caller with no host therefore keeps its paths.
• Who reads the labels. Server-side, no module reads a section's label or note except the parity test: no prompt and no wire field carries them.

tests/test_doxbench_scope.py

• The browser harness copies display.js beside the two models and writes a package.json with "type": "module" (DJ). It installs the display through setDisplay(readDisplay(...)), as the shell does at mount.
• The parity test hands the browser the served display (VR).
• Two new cases, because the parity test alone cannot catch one thing. The composite serves openDox's neutral stage words for these three roles, since openXdox overlays only completion and the values. So parity cannot tell a server that reads the facet from one that always speaks neutral words. The new cases can:
  • A host that declares its own stage words. It declares openxFactory's pre-S7 words. The server labels read "cluster documents", "topic folder documents" and so on, and match the browser byte for byte.
  • No host. The server reads the neutral words, which equal the browser's with no facet.

tests/test_canvas.py

• _run_node copies display.js and writes the package.json (DJ). It hands the harness the served display, so every canvas derivation runs under the words the page is served (VR).
• The two gap pins now read the kind canvas-model.js emits under every host, unsupported-candidate, where they read unsupported-possible before (R9-R1).
• test_js_and_python_draft_constants_agree also cross-checks the seed state of a new draft. Both halves read it off one facet: the composer plan's state against canvas_drafts.register_state("captured"). This assertion is what makes the served-display hand-off observable (mutant H4 below).

tests/test_explorer_viewer.py

• DJ: each explorer harness copies display.js and writes the package.json.
• VR: _EXPLORER_HARNESS is handed the served display.
• R9-R1, the served words. The two label pins read the words the page is served, from the composite's artifact axis:
  • groups proposal.md and tasks.md;
  • classification proposal.md, design.md, tasks.md, Spec deltas, Supporting docs, other.
• CP: the import-free pin, respelled to S7's design:
  • no import names the viewer;
  • the only module imported is ./display.js;
  • import statements spread over several lines are read whole.
• CP: the wheel pin, respelled to opts.wheelKey === SOURCE, with SOURCE held to be STAGE_ROLES[0].

The falsifier (composed, each file alone; quoted)

The composed form, R1Q23 (a):

• openxFactory 92010d3e (tests/composed_host_pin.yaml), with openDox, openXdox and openXwallet initialized recursively, and its scripts/ on PYTHONPATH;
• openDox installed from openDox-code main 84f8ed83 (force-reinstalled);
• the colour variables unset and TERM=dumb;
• per file: python -m pytest -q -p no:cacheprovider -rfE --junitxml=<file>.xml tests/<file>.py < /dev/null.

T022's scripts/composed_placements.py (taken from #44's tree) refuses to place in these checkouts: its .gitignore entry is not on main until #41 lands. None of these files reads the runbook it places. Every checkout read clean after every run.

Measured over real main (e2340bb6, carrying T095 #46 and T026 #43): the base is main itself, and the head is this PR's c935f2a0 (this branch merged with main).

run	test_canvas.py	test_explorer_viewer.py	test_doxbench_scope.py	named reds
base: main e2340bb6	10 passed, 11 failed	15 passed, 9 failed	34 passed, 1 failed	21
head: c935f2a0	21 passed, 0 failed	24 passed, 0 failed	37 passed, 0 failed	0

Before T095 landed, the same 21 were measured at a local integration that was never pushed (main f70bc9c3 + #46 c2192e3 + this branch): 21 red at its base and 0 at its head. At main d45a0939, before T095 existed, these files read 19, 9 and 1 red.

The 21 nodes that flipped:

• canvas (11): test_member_pane_is_exactly_the_topics_document_edges, test_downstream_artifacts_are_in_the_lineage_strip_never_members, test_evidence_board_carries_section_ref_and_passage_hash, test_missing_cluster_degrades_to_null, test_gap_unclaimed_member, test_gap_unsupported_possible, test_gap_both_kinds_on_one_cluster, test_option_set_grouping, test_js_and_python_draft_constants_agree, test_esc_escapes_markup, test_model_passes_prose_through_as_data_not_html;
• explorer (9): test_explorer_js_exists_and_is_importable_standalone, test_staged_tile_lists_the_topic_folder_with_headers, test_aggregate_folder_entry_keeps_its_member_source_key, test_aggregate_folder_entry_disambiguates_same_repository_refs, test_aggregate_wheel_navigation_threads_the_owning_member_key, test_proposal_tile_groups_proposal_and_tasks, test_realized_tile_lists_the_archived_change_folder, test_unresolvable_tile_and_kind_degrade_to_null, test_change_file_classification;
• scope (1): test_server_and_actual_browser_projection_are_byte_shape_equivalent.

Two nodes are new: test_the_server_labels_follow_the_facet_a_host_declares and test_with_no_host_registered_the_labels_are_openDoxs_neutral_words. No node regressed.

The unchanged set was compared node for node, base against head over main: 974 nodes, 10 red at both, 0 nodes changed outcome.

set	base	head
12.5's 16 governed suites	832 passed, 10 failed (pre-existing: session_runbook 9 and session_notebook 1, which read the runbook T022 places)	identical, file for file
F5.2's 5 suites	123 passed	identical, file for file
test_column_contributions_governed.py	9 passed	9 passed

Mutants

Each mutant was applied at the integration (main f70bc9c3 + T095 c2192e3 + this branch), its file run alone composed, and the file restored byte for byte. 11 of 12 killed:

mutant	site	killed by
S1	src: ignores the host's facet (always neutral)	test_the_server_labels_follow_the_facet_a_host_declares
S2	src: no-host fallback dropped	test_with_no_host_registered_the_labels_are_openDoxs_neutral_words
S3	src: members label hard-coded "cluster documents"	parity (served) + no-host
S4	src: folder label one/many swapped	parity + declaring host + no-host
S5	src: neighbourhood note drops a facet word	declaring host
H1	scope harness: no display.js	all three browser cases
H2	scope harness: browser ignores the display	declaring host
H3	canvas: no display.js	the 11 canvas Node cases
H4	canvas: neutral display instead of served	test_js_and_python_draft_constants_agree
H5	explorer: no display.js	7 explorer Node cases
H6	explorer: neutral display instead of served	test_proposal_tile_groups_proposal_and_tasks, test_change_file_classification
H7	scope harness: no package.json	SURVIVES (37 passed)

H7 is an equivalent mutant on the measured runtime. Node v24 detects module syntax in a .js file with no "type" declared, so package.json changes nothing there. It is kept for runtimes without that detection, as #43's harness keeps it. T097 measured the same survivor.

The lone CI form (validate.yml's exact step, at c935f2a0)

python -m pytest -q --junitxml=pytest-report.xml --deselect "$LEFT_OUT", run locally, read:

SELECTED=1165 PASSED=1161 SKIPPED=4 FAILURES=0 ERRORS=0   (floors MIN_SELECTED=1099 MIN_PASSED=1095 EXPECT_SKIPPED=4)
1161 passed, 4 skipped, 1 deselected in 169.43s

• tests/test_declared_exclusion.py run alone: rc 0. Each of the three files still fails lone on its declared reason (doc_health, at collection).
• The floors are untouched; only T029 raises them.

CI and Copilot at c935f2a0

• validate: success, run 37548181742. The triple is selected=1165 passed=1161 skipped=4 failures=0 errors=0, with pins selected>=1099 (margin 66), passed>=1095 (margin 66) and skipped==4.
• SonarCloud Code Analysis: success.
• Copilot, review 5435780547 at c935f2a0:
  • "Needs a closer look", Findings: None. Its stated reason is that "final composed validation after integration remains unverified". This PR's CI cannot run the composed form until composed.yml lands (T029 DRAFT: T029, U-9: the composed run, permanent: composed.yml and its openxFactory pin (plan 038) #44, then T073 DRAFT: T073: the composed declarations (ARC-Q2 (a)): every declared entry runs composed, the rail files get the host's rail (plan 038) #45). The composed validation over real main e2340bb6 is the local run quoted above: 21 to 0 red, and the unchanged set identical node for node.
  • The earlier review at 588d4c15 (5435186246) gave the same verdict, with no findings, for the then-unmerged T095: the composed harness for the declared integration tests (plan 038, T094's R2-INV-R9) #46.
  • 0 review threads, 0 unresolved.

Files changed

• src/openxdox/doxbench_scope.py
• tests/test_canvas.py
• tests/test_explorer_viewer.py
• tests/test_doxbench_scope.py

No protected suite is edited: none of the four is in 12.5's 16 or F5.2's 7. No allow-list entry is needed, and tests/declared_exclusion.yaml is untouched.

Not done

• Composed CI: composed.yml is not on main yet (T029 DRAFT: T029, U-9: the composed run, permanent: composed.yml and its openxFactory pin (plan 038) #44, then T073 DRAFT: T073: the composed declarations (ARC-Q2 (a)): every declared entry runs composed, the rail files get the host's rail (plan 038) #45), so this PR's CI is validate and SonarCloud. The composed numbers above are local runs, quoted.

🤖 Generated with Claude Code

Summary by Sourcery

Use the served display facet consistently across server and browser projections so labels, classifications, and derived values remain aligned.

New Features:

• Make scope section labels and notes follow the served display facet, including host-specific vocabulary and neutral fallback words when no host is registered.
• Pass the served display facet into canvas, explorer, and scope browser-model harnesses so projections use the same vocabulary as the server.

Bug Fixes:

• Align server and browser projections with the served facet, including updated gap kinds, artifact labels, and file classifications.

Enhancements:

• Update standalone JavaScript harnesses to include the shared display module and module configuration.
• Strengthen parity and facet-specific tests for host-declared and hostless display vocabulary.

Tests:

• Add coverage for served-display parity, host-specific scope labels, neutral no-host labels, and shared canvas draft state.

[sourcery-ai]

Reviewer's Guide

This draft aligns server scope metadata and browser projections around the served display facet, lazily resolving host vocabulary with neutral fallback, and updates the Node harnesses and assertions to validate display propagation, role-based labels, and parity across hosted and hostless cases.

Sequence diagram for served display vocabulary during scope resolution

sequenceDiagram
    participant Caller
    participant Scope as doxbench_scope
    participant Registry as domain_profile
    participant Display as display_profile

    Caller->>Scope: resolve_scope(...)
    Scope->>Registry: is_registered()
    alt host registered
        Scope->>Display: host_display()
        Display-->>Scope: host display facet
    else no host
        Scope->>Display: display_manifest(None)
        Display-->>Scope: neutral display facet
    end
    Scope->>Display: display_manifest(host display)
    Display-->>Scope: served display
    Scope->>Scope: _section_meta(served display)
    Scope-->>Caller: scope sections with facet-derived labels

Loading

File-Level Changes

Change	Details	Files
Derive server-side scope section metadata from the served display facet instead of fixed domain-specific wording.	Build labels and notes from source, grouping, and selection stage words while preserving section keys and flags. Resolve the display at scope-resolution time through the registered host, with neutral-word fallback when no host exists. Thread the generated metadata through all scope section builders.	src/openxdox/doxbench_scope.py
Update browser-based harnesses to execute display-aware JavaScript with the same served facet as the page.	Copy display.js beside each tested model and add module-package setup. Pass the served display into canvas, explorer, and scope projection derivations. Update assertions for facet-derived labels, role-based gap kinds, wheel keys, and draft-state parity.	tests/test_canvas.py tests/test_explorer_viewer.py tests/test_doxbench_scope.py
Expand scope parity coverage to validate host-specific vocabulary and no-host behavior.	Add a custom host facet case that checks server/browser byte parity and legacy host wording. Add a no-host case that verifies neutral wording without rejecting scope resolution. Retain import-isolation coverage while allowing the display.js dependency.	tests/test_doxbench_scope.py

---

Tips and commands

Interacting with Sourcery

• Trigger a new review: Comment @sourcery-ai review on the pull request.
• Continue discussions: Reply directly to Sourcery's review comments.
• Generate a GitHub issue from a review comment: Ask Sourcery to create an
  issue from a review comment by replying to it. You can also reply to a
  review comment with @sourcery-ai issue to create an issue from it.
• Generate a pull request title: Write @sourcery-ai anywhere in the pull
  request title to generate a title at any time. You can also comment
  @sourcery-ai title on the pull request to (re-)generate the title at any time.
• Generate a pull request summary: Write @sourcery-ai summary anywhere in
  the pull request body to generate a PR summary at any time exactly where you
  want it. You can also comment @sourcery-ai summary on the pull request to
  (re-)generate the summary at any time.
• Generate reviewer's guide: Comment @sourcery-ai guide on the pull
  request to (re-)generate the reviewer's guide at any time.
• Resolve all Sourcery comments: Comment @sourcery-ai resolve on the
  pull request to resolve all Sourcery comments. Useful if you've already
  addressed all the comments and don't want to see them anymore.
• Dismiss all Sourcery reviews: Comment @sourcery-ai dismiss on the pull
  request to dismiss all existing Sourcery reviews. Especially useful if you
  want to start fresh with a new review - don't forget to comment
  @sourcery-ai review to trigger a new review!

Customizing Your Experience

Access your dashboard to:

• Enable or disable review features such as the Sourcery-generated pull request
  summary, the reviewer's guide, and others.
• Change the review language.
• Add, remove or edit custom review instructions.
• Adjust other review settings.

Getting Help

• Contact our support team for questions or feedback.
• Visit our documentation for detailed guides and information.
• Keep in touch with the Sourcery team by following us on X/Twitter, LinkedIn or GitHub.

[Copilot]

Copilot review overview

🔵 Needs a closer look

Required prerequisite #46 remains unmerged, and composed validation must be repeated after integration.

Review effort: Balanced
Findings: None

What changed in this PR

Aligns openXdox’s scope metadata and browser test harnesses with the served display vocabulary.

Changes:

• Derives scope labels from host display settings, with neutral fallback.
• Supplies display dependencies and vocabulary to browser harnesses.
• Updates terminology assertions and adds host-specific and no-host parity coverage.

File	Description
tests/​test_explorer_viewer.py	Updates display-aware harnesses and dependency assertions.
tests/​test_doxbench_scope.py	Adds host-specific and neutral vocabulary parity tests.
tests/​test_canvas.py	Supplies served display and checks draft-state parity.
src/​openxdox/​doxbench_scope.py	Derives section metadata from the display facet.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[Copilot]

Copilot review overview

🔵 Needs a closer look

Prerequisite #46 has landed, but final composed validation after integration remains unverified.

Review effort: Balanced
Findings: None

[sourcery-ai]

Sorry @brettheap, you've used your own review budget of 250,000 diff characters for the last 7 days.

You can request another review in 21 hours and 53 minutes by commenting @sourcery-ai review. Upgrade to get a review now.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[Copilot]

Copilot review overview

🟢 Approval recommended

No blocking issues remain; the targeted checks support the existing coverage, and authorization paths are unchanged.

Review effort: Balanced
Findings: None

[brettheap]

READY at 55dc1ee

Lane: openxfactory-4 (openXfactory-4-openDox_extraction)

Holder, automatic READY on Brett's word "land lane 3's PRs when green" (2026-10-06). At the head: CLEAN, rollup SUCCESS, nothing pending, 0 unresolved threads, and Copilot at 55dc1ee7 reads "🟢 Approval recommended". The per-repo lander checks closing keywords and lands it.

[brettheap]

Lane: openxfactory-4 (openXfactory-4-openDox_extraction)

LANDED — lane openxfactory-4, 2026-10-07T00:04:10Z, PR #51 → fc1115d (opensoft/openXdox-code main; plain gate)

Brett: plan 038 RULED incl. lane openxfactory-4 lands every PR (#656 6013547504); Start release 2 (#656 6001702967)