Publish the articles-v11 reference texts for the 530-task revision

Add the articles suite under definitions/materials/primary-content-20261103 from the
hash-matched articles-v11 dataset (PR #5), bind its resolved cases in the population,
and record the article control results. All 530 selected tasks are now scorable from
the published archive.

README.md

@@ -98,8 +98,8 @@ contract**, using the fixed 530-task selection and fresh matched runs after adop
 It does not certify the old report or establish that all similar requirements in
 the catalog have been corrected. The 66 earlier cases have source-context review
 records, which disclose partial observations and unavailable historical captures.
-The three later Xueqiu cases get their controls under the unified rule when the
-article reference inputs are available.
+The three later Xueqiu cases have no full-page capture among the bound inputs (see
+the review notes).
 Review here means assistant-led source inspection and automated checks, not
 independent human approval. Raw captures are external, so these checks cannot be
 replayed using the public repository alone. Independent sampling of the removals
@@ -235,12 +235,10 @@ Browser Eval downloads the dataset revision selected by its [data lock](https://
 
 `definitions/materials/primary-content-20261103/<suite>/dataset.json` contains the
 complete scorable tasks, with `primary-content-20261103` already applied, for
-`webfetch`, `dynamic-content`, `popular-sites` and `access-restrictions`. It
-covers 430 of the 530 selected tasks. Each directory's `source-inputs.json` and
+`webfetch`, `dynamic-content`, `popular-sites`, `access-restrictions` and `articles`.
+Together they cover all 530 selected tasks. Each directory's `source-inputs.json` and
 `dataset-catalog.json` bind the file by SHA-256. The selection's
 `frozen_task_bindings[<suite>].exported_cases_sha256` binds its resolved cases.
-The `articles` reference texts are pending, so that suite is not scorable from
-this download yet.
 
 Each file is the frozen dataset bound by the suite's `definitions/evaluation/.../source-inputs.json`.
 Its hashes were verified, and it reproduces every public task row. The Browser Eval

THIRD_PARTY.md

@@ -4,9 +4,9 @@ The Apache-2.0 license covers project-owned code, documentation and task require
 
 ## Reference texts included for evaluation
 
-`definitions/materials/primary-content-20261103/` includes the reference texts that scoring needs for four suites (webfetch, dynamic-content, popular-sites and access-restrictions): main-content references, reference body regions and article endings, kept in their original language and wording. They are included only so the benchmark can be scored from a clean download. The Apache-2.0 license does not cover them. Each text keeps the terms of its source website, and inclusion does not grant any further right to reuse it.
+`definitions/materials/primary-content-20261103/` includes the reference texts that scoring needs for all five suites (webfetch, dynamic-content, popular-sites, access-restrictions and articles): main-content references, reference body regions and article endings, kept in their original language and wording. They are included only so the benchmark can be scored from a clean download. The Apache-2.0 license does not cover them. Each text keeps the terms of its source website, and inclusion does not grant any further right to reuse it.
 
-Page captures, screenshots, document and media files, source observation files and measured provider outputs are not included. Article reference texts (`articles-v11`) will be added in a later revision.
+Page captures, screenshots, document and media files, source observation files and measured provider outputs are not included.
 
 If you own rights in an included text and want it removed or changed, open a discussion in this dataset repository's Community tab. Give the task ID and source URL. We will remove or replace the text and record the change in a new revision. Earlier revisions stay in the repository history.
 

definitions/annotation-revisions/primary-content-20261103-rationale.json

@@ -6495,7 +6495,7 @@
         }
       },
       "rule": "xueqiu-shell-v1",
-      "pending": "Full-answer and error-page controls for the unified Xueqiu rule run during `export --revision` once the articles-v11 reference inputs are available."
+      "pending": "Full-answer control uses the declared review evidence (export --review-source); with the frozen dataset alone the error page fails."
     },
     {
       "suite": "articles",
@@ -6704,7 +6704,7 @@
         }
       },
       "rule": "xueqiu-shell-v1",
-      "pending": "Full-answer and error-page controls for the unified Xueqiu rule run during `export --revision` once the articles-v11 reference inputs are available."
+      "pending": "Full-answer control uses the declared review evidence (export --review-source); with the frozen dataset alone the error page fails."
     },
     {
       "suite": "dynamic-content",
@@ -10550,7 +10550,7 @@
         }
       },
       "rule": "xueqiu-shell-v1",
-      "pending": "Full-answer and error-page controls for the unified Xueqiu rule run during `export --revision` once the articles-v11 reference inputs are available."
+      "pending": "Full-answer control uses the declared review evidence (export --review-source); with the frozen dataset alone the error page fails."
     },
     {
       "suite": "articles",
@@ -10956,7 +10956,7 @@
         "status": "ran against the earlier, narrower Xueqiu edit"
       },
       "rule": "xueqiu-shell-v1",
-      "pending": "Full-answer and error-page controls for the unified Xueqiu rule run during `export --revision` once the articles-v11 reference inputs are available."
+      "pending": "Full-answer control needs a full-page capture; the bound body reference satisfies the body regions but not the retained author, time, risk-notice and comment-label requirements. The error page fails."
     },
     {
       "suite": "articles",
@@ -11130,7 +11130,7 @@
         "status": "ran against the earlier, narrower Xueqiu edit"
       },
       "rule": "xueqiu-shell-v1",
-      "pending": "Full-answer and error-page controls for the unified Xueqiu rule run during `export --revision` once the articles-v11 reference inputs are available."
+      "pending": "Full-answer control needs a full-page capture; the bound body reference satisfies the body regions but not the retained author, time, risk-notice and comment-label requirements. The error page fails."
     },
     {
       "suite": "articles",
@@ -11352,7 +11352,7 @@
         "status": "ran against the earlier, narrower Xueqiu edit"
       },
       "rule": "xueqiu-shell-v1",
-      "pending": "Full-answer and error-page controls for the unified Xueqiu rule run during `export --revision` once the articles-v11 reference inputs are available."
+      "pending": "Full-answer control needs a full-page capture; the bound body reference satisfies the body regions but not the retained author, time, risk-notice and comment-label requirements. The error page fails."
     }
   ],
   "retained": [

definitions/annotation-revisions/primary-content-20261103-review.md

@@ -48,14 +48,16 @@ should sample the removals.
    own series, author, date, topic and related-content requirements, so neither
    changes here. article-091 also keeps its comment-moderation notice as part of the
    displayed comment area; article-064 never required such a notice.
-8. **Pending controls.** The saved-answer replays of article-044, -102 and -104 ran
-   against the earlier, narrower Xueqiu edit. Complete-answer and error-page controls
-   for the unified rule run during `export --revision` once the `articles-v11`
-   reference inputs are available.
+8. **Article controls.** With the frozen `articles-v11` dataset, every revised article
+   task rejects the access-denied page. Article-044, -102 and -104 are bound only to
+   their body reference. That reference satisfies their body regions, but it lacks
+   the page-level requirements the rule keeps: author, time, risk notice and comment
+   labels. Their full-answer controls therefore need a full-page capture, which the
+   bound inputs do not contain. The earlier saved-answer replays ran against the
+   narrower Xueqiu edit.
 
 ## Reference texts
 
 `definitions/materials/primary-content-20261103/` publishes the resolved, scorable
-datasets for webfetch, dynamic-content, popular-sites and access-restrictions. Each one
-is the frozen dataset bound by that suite's `source-inputs.json`, with this revision
-applied. The articles suite will be added when its frozen dataset is supplied.
+datasets for all five suites. Each one is the frozen dataset bound by that suite's
+`source-inputs.json`, with this revision applied.

definitions/annotation-revisions/primary-content-20261103.md

@@ -88,9 +88,8 @@ Successful export does not certify absent source evidence.
 
 The public archive contains task definitions, rule changes and, under
 `definitions/materials/primary-content-20261103/`, the resolved scorable datasets
-for 430 of the 530 tasks. Their reference texts keep their source terms (see
-THIRD_PARTY.md). The 100 article tasks remain unscorable from the archive until the
-`articles-v11` frozen dataset is added.
+for all 530 tasks. Their reference texts keep their source terms (see
+THIRD_PARTY.md).
 
 After source files are uploaded, run `uv run --frozen python
 tools/rebuild_publication.py --source-revision <uploaded-full-HF-commit>` and

definitions/materials/primary-content-20261103/articles/source-inputs.json

@@ -0,0 +1,13 @@
+{
+  "schema_version": 2,
+  "kind": "frozen_reference_inputs",
+  "id": "primary-content-20261103-articles",
+  "catalog": "dataset-catalog.json",
+  "catalog_sha256": "e3316f875956276137569fb0c7334fd2f25485dd478ee772c310e3bd603a99c0",
+  "dataset": "dataset.json",
+  "dataset_sha256": "c25cbdcd338b3d999fbb6ad5198edfa61d05d4487f0eab1f3326c3e35b00ea39",
+  "files": {
+    "dataset.json": "c25cbdcd338b3d999fbb6ad5198edfa61d05d4487f0eab1f3326c3e35b00ea39"
+  },
+  "trees": {}
+}

definitions/populations/primary-content-20261103.json

@@ -34,7 +34,8 @@
     },
     "articles": {
       "path": "definitions/evaluation/articles/tasks.json",
-      "sha256": "171241ea0bf3cb3f9482de14a7201885733fa717af56d7ea2b28d86a4516a3bb"
+      "sha256": "171241ea0bf3cb3f9482de14a7201885733fa717af56d7ea2b28d86a4516a3bb",
+      "exported_cases_sha256": "9e8bf7fa1a9d4eba423c32b47582d19d0129279823a3ad1143e161e0a7ce85c3"
     }
   },
   "cases": [

tools/test_materials.py

@@ -7,7 +7,7 @@ import unittest
 ROOT = Path(__file__).resolve().parents[1]
 REVISION = 'primary-content-20261103'
 MATERIALS = ROOT / 'definitions/materials' / REVISION
-PUBLISHED = {'webfetch', 'dynamic-content', 'popular-sites', 'access-restrictions'}
+PUBLISHED = {'webfetch', 'dynamic-content', 'popular-sites', 'access-restrictions', 'articles'}
 
 
 def canonical(value):
@@ -22,8 +22,7 @@ class MaterialsTests(unittest.TestCase):
 
     def test_inventory(self):
         self.assertEqual({p.name for p in MATERIALS.iterdir()}, PUBLISHED)
-        bindings = self.population['frozen_task_bindings']
-        self.assertNotIn('exported_cases_sha256', bindings['articles'])  # pending reference texts
+        self.assertEqual(set(self.population['frozen_task_bindings']), PUBLISHED)
 
     def test_manifests_bind_dataset_catalog_and_revision(self):
         for suite in PUBLISHED:
@@ -55,5 +54,8 @@ class MaterialsTests(unittest.TestCase):
         for suite in PUBLISHED:
             dataset = json.loads((MATERIALS / suite / 'dataset.json').read_text())
             for case in dataset['cases']:
-                for region in (case.get('body_requirements') or {}).get('regions', []):
+                body = case.get('body_requirements') or {}
+                for region in body.get('regions', []):
                     self.assertIn('text', region, (suite, case['id']))
+                if 'ending' in body:
+                    self.assertTrue(body['ending'].get('text'), (suite, case['id']))