fix(context): drop archived pages from context packs

src/vouch/context.py

@@ -586,6 +586,22 @@ def search_kb(
     )
 
 
+def _page_is_live(store: KBStore, page_id: str) -> bool:
+    """False for an archived page, or one whose yaml is gone.
+
+    Shared by ``kb.search``'s hit filter and both context-pack builders. The
+    claim half of this predicate is inlined at each call site because those
+    callers need the fetched claim anyway (citations, origin tags); pages are
+    only ever tested, so the check lives here once — keeping it in three
+    places is what let ``kb.context`` keep serving archived pages after #581
+    fixed ``kb.search``.
+    """
+    try:
+        return store.get_page(page_id).status is not PageStatus.ARCHIVED
+    except ArtifactNotFoundError:
+        return False
+
+
 def _filter_live_hits(
     store: KBStore,
     hits: list[tuple[str, str, str, float]],
@@ -607,13 +623,8 @@ def _filter_live_hits(
                 continue
             if claim.status in _RETRACTED_CLAIM_STATUSES:
                 continue
-        elif kind == "page":
-            try:
-                page = store.get_page(artifact_id)
-            except ArtifactNotFoundError:
-                continue
-            if page.status is PageStatus.ARCHIVED:
-                continue
+        elif kind == "page" and not _page_is_live(store, artifact_id):
+            continue
         kept.append((kind, artifact_id, summary, score))
         if limit is not None and len(kept) >= limit:
             break
@@ -674,6 +685,8 @@ def _append_graph_neighbors(
             if claim.status in _RETRACTED_CLAIM_STATUSES:
                 continue
             cites = list(claim.evidence)
+        elif kind == "page" and not _page_is_live(store, nid):
+            continue
         via = node.get("via", "")
         parent_score = seed_scores.get(via, 0.5)
         distance = int(node.get("distance", 1))
@@ -789,6 +802,10 @@ def build_context_pack(
                 continue
             cites = list(claim.evidence)
             origin = _origin_from_tags(claim.tags)
+        elif kind == "page" and not _page_is_live(store, hid):
+            # Archiving a page must remove it from recall, not just from
+            # kb.search — this is the surface that seeds agent context.
+            continue
         summary = _enrich_summary(store, kind, hid, summary)
         items.append(
             ContextItem(

tests/test_context.py

@@ -172,6 +172,49 @@ def test_context_pack_excludes_archived_claims(store: KBStore) -> None:
     assert not any(it["id"] == "c1" for it in pack["items"]), pack
 
 
+def test_context_pack_excludes_archived_pages(store: KBStore) -> None:
+    """The mirror of #581 on the recall side: kb.search learned to drop
+    archived pages, build_context_pack never did — so archiving a page hid
+    it from search while it kept being injected into every context pack."""
+    from vouch.models import Page, PageStatus, PageType
+
+    src = store.put_source(b"e")
+    store.put_page(Page(
+        id="p-live", title="mongodb ops", body="live page about mongodb",
+        type=PageType.CONCEPT, sources=[src.id],
+    ))
+    store.put_page(Page(
+        id="p-arch", title="mongodb old", body="archived mongodb notes",
+        type=PageType.CONCEPT, status=PageStatus.ARCHIVED, sources=[src.id],
+    ))
+    health.rebuild_index(store)
+
+    pack = context.build_context_pack(store, query="mongodb", limit=10)
+    ids = {item["id"] for item in pack["items"]}
+
+    assert "p-live" in ids, pack
+    assert "p-arch" not in ids, pack
+
+
+def test_page_is_live_rejects_archived_and_missing(store: KBStore) -> None:
+    """The shared predicate behind both context builders and kb.search.
+
+    The graph-expansion path carried the same gap as the main loop and now
+    routes through this, so pinning its contract pins both call sites.
+    """
+    from vouch.models import Page, PageStatus, PageType
+
+    src = store.put_source(b"e")
+    store.put_page(Page(
+        id="p-arch", title="mongodb old", body="archived mongodb notes",
+        type=PageType.CONCEPT, status=PageStatus.ARCHIVED, sources=[src.id],
+    ))
+    health.rebuild_index(store)
+
+    assert context._page_is_live(store, "p-arch") is False
+    assert context._page_is_live(store, "p-missing") is False
+
+
 def test_search_kb_excludes_retracted_claims(store: KBStore) -> None:
     """Regression for #581: search_kb must apply the same retracted-status
     filter as build_context_pack — otherwise kb.search leaks archived /