diff --git a/README.md b/README.md
index a27508a6..daaeaf4e 100644
--- a/README.md
+++ b/README.md
@@ -235,14 +235,24 @@ provider-backed ELF evidence was required.
   ambiguous, stale, and superseded markers without introducing a separate graph
   database or replacing source evidence.
 - Recall/debug panel after XY-1022: the June 20 follow-up adds
-	  `elf.recall_debug_panel/v1` through service, HTTP, and MCP readback. The panel
-	  groups Memory Note trace selected rows and retained dropped replay candidates,
-	  Source Library document candidates, Knowledge Workspace page snippets, graph facts,
-	  and Dreaming proposals with
-	  authority layer, freshness state, source refs, stage reason, evidence class, and
-	  replay command. Missing anchors remain explicit `not_requested` layers, so the
-	  panel improves debug ergonomics without turning untested or blocked layers into
-	  pass claims.
+  `elf.recall_debug_panel/v1` through service, HTTP, and MCP readback. The panel
+  groups Memory Note trace selected rows and retained dropped replay candidates,
+  Source Library document candidates, Knowledge Workspace page snippets, graph facts,
+  and Dreaming proposals with
+  authority layer, freshness state, source refs, stage reason, evidence class, and
+  replay command. Missing anchors remain explicit `not_requested` layers, so the
+  panel improves debug ergonomics without turning untested or blocked layers into
+  pass claims.
+- Agent Knowledge OS closeout after XY-1023: the June 20 closeout report publishes
+  the full product/scenario matrix for 19 tracked products and six Agent Knowledge OS
+  layers, after rerunning `cargo make real-world-memory` at 62 jobs, 55 pass,
+  0 wrong_result, and 7 typed blockers. ELF is the strongest measured integrated
+  Agent Knowledge OS product because all six ELF-owned layers have checked-in
+  evidence, but the report preserves qmd
+  retrieval/debug ergonomics, OpenViking trajectory, mem0/OpenMemory history and
+  UI/export, Letta core/archive, graph/RAG temporal-citation, agentmemory/claude-mem
+  capture/viewer, and VectifyAI PageIndex/OpenKB long-document knowledge-library
+  advantages as optimization inputs rather than false pass claims.
 - Operator-approved public-proxy addendum after XY-930: the June 19 follow-up runs
   `cargo make baseline-production-private-addendum` with a simulated/public-proxy
   production corpus manifest approved for this stage. The run records 12 documents,
@@ -373,6 +383,7 @@ Detailed evidence and interpretation:
 - [Graph Topic-Map Report - June 20, 2026](docs/evidence/benchmarking/2026-06-20-graph-topic-map-report.md)
 - [Knowledge Workspace Version-Diff Report - June 20, 2026](docs/evidence/benchmarking/2026-06-20-knowledge-workspace-version-diff-report.md)
 - [Live Knowledge-Page Rebuild/Lint Report - June 20, 2026](docs/evidence/benchmarking/2026-06-20-live-knowledge-page-rebuild-lint-report.md)
+- [Agent Knowledge OS Closeout Benchmark Report - June 20, 2026](docs/evidence/benchmarking/2026-06-20-agent-knowledge-os-closeout-benchmark-report.md)
 - [Live Baseline Benchmark Runbook](docs/runbook/benchmarking/live_baseline_benchmark.md)
 - [Single-User Production Runbook](docs/runbook/single_user_production.md)
 - Benchmark contract:
@@ -474,9 +485,9 @@ Detailed comparison, mechanism-level analysis, and source map:
 - [Dreaming Product Surface Follow-Up Research](docs/research/dreaming_product_surface_followup.md)
 
 Latest real-world benchmark report: June 20, 2026. Latest external research refresh:
-June 11, 2026; June 20 adds the Graph Topic-Map Report - June 20, 2026,
-Knowledge Workspace Version-Diff Report - June 20, 2026, and the Live
-Knowledge-Page Rebuild/Lint Report - June 20, 2026 after the June 19
+June 11, 2026; June 20 adds the Agent Knowledge OS Closeout Benchmark Report,
+the Graph Topic-Map Report - June 20, 2026, Knowledge Workspace Version-Diff
+Report - June 20, 2026, and the Live Knowledge-Page Rebuild/Lint Report - June 20, 2026 after the June 19
 XY-930 operator-approved public-proxy production addendum and service-native Dreaming
 readback, the qmd debug-ergonomics Dreaming retest, the June 17 competitor-strength
 closeout, and the June 16 temporal reconciliation, live consolidation self-check,
diff --git a/apps/elf-eval/fixtures/report_snapshots/2026-06-20-agent-knowledge-os-closeout-benchmark-report.json b/apps/elf-eval/fixtures/report_snapshots/2026-06-20-agent-knowledge-os-closeout-benchmark-report.json
new file mode 100644
index 00000000..5cc01771
--- /dev/null
+++ b/apps/elf-eval/fixtures/report_snapshots/2026-06-20-agent-knowledge-os-closeout-benchmark-report.json
@@ -0,0 +1,521 @@
+{
+  "schema": "elf.agent_knowledge_os_closeout_benchmark_report/v1",
+  "authority": "XY-1023",
+  "source_contract": "elf-agent-knowledge-os-2026-06-20",
+  "source_issue": "XY-286",
+  "generated_at": "2026-06-20T00:00:00Z",
+  "report_markdown": "docs/evidence/benchmarking/2026-06-20-agent-knowledge-os-closeout-benchmark-report.md",
+  "all_project_fixture_rerun": {
+    "command": "cargo make real-world-memory",
+    "status": "pass",
+    "run_id": "real-world-memory",
+    "job_count": 62,
+    "encoded_suite_count": 17,
+    "pass": 55,
+    "wrong_result": 0,
+    "incomplete": 0,
+    "blocked": 7,
+    "not_encoded": 0,
+    "evidence_coverage": 1.0,
+    "source_ref_coverage": 1.0,
+    "quote_coverage": 1.0,
+    "mean_score": 0.887
+  },
+  "summary": {
+    "strongest_measured_integrated_product": "ELF integrated Agent Knowledge OS",
+    "strongest_product_qualification": "ELF is the strongest measured integrated product in the checked-in Agent Knowledge OS matrix because all six ELF-owned layers now have source-linked readback, tests, and docs evidence. This is not a broad claim that ELF beats every competitor on every competitor-owned specialty.",
+    "complete_same_corpus_product_count": 1,
+    "product_count": 19,
+    "scenario_count": 6,
+    "matrix_cell_count": 114,
+    "coverage_statement": "Only ELF has complete same-repo coverage across all six Agent Knowledge OS layers. qmd, agentmemory, OpenViking, mem0/OpenMemory, claude-mem, memsearch, Letta, graph/RAG projects, llm-wiki, gbrain, PageIndex, and OpenKB have partial, blocked, or reference-only coverage depending on the scenario.",
+    "not_every_product_has_complete_live_coverage": true,
+    "evidence_class_counts": {
+      "pass": 9,
+      "wrong_result": 7,
+      "incomplete": 6,
+      "blocked": 14,
+      "not_tested": 78
+    }
+  },
+  "supported_scenarios": [
+    {
+      "id": "source_library_ingest_hydration",
+      "layer": "Source Library",
+      "elf_evidence": "docs/evidence/benchmarking/2026-06-20-recall-debug-panel-report.md and source-library fixture coverage",
+      "completion_rule": "Captured sources remain documents with source refs, hydration, search, and bounded replay; they are not silently converted into authoritative memory notes."
+    },
+    {
+      "id": "memory_authority_history_read_profiles",
+      "layer": "Memory Authority",
+      "elf_evidence": "docs/evidence/benchmarking/2026-06-16-live-temporal-reconciliation-report.md and service-native Dreaming readback",
+      "completion_rule": "Current, historical, tombstoned, stale, superseded, selected, and dropped evidence stays explicit under read-profile policy."
+    },
+    {
+      "id": "knowledge_workspace_pages",
+      "layer": "Knowledge Workspace",
+      "elf_evidence": "docs/evidence/benchmarking/2026-06-20-live-knowledge-page-rebuild-lint-report.md and 2026-06-20-knowledge-workspace-version-diff-report.md",
+      "completion_rule": "Knowledge pages are rebuildable derived artifacts with source refs, lint, version diffs, and no hidden source-of-truth mutation."
+    },
+    {
+      "id": "temporal_topic_graph_lite",
+      "layer": "Graph-lite Facts",
+      "elf_evidence": "docs/evidence/benchmarking/2026-06-20-graph-topic-map-report.md",
+      "completion_rule": "Graph facts are typed Postgres readback over sourced, inferred, ambiguous, stale, and superseded markers, not a second graph database authority."
+    },
+    {
+      "id": "dreaming_review_queue",
+      "layer": "Dreaming Review Queue",
+      "elf_evidence": "docs/evidence/benchmarking/2026-06-20-dreaming-review-queue-report.md",
+      "completion_rule": "Background proposals stay reviewable with source refs, affected refs, lint, diff, policy, and review audit before high-impact mutation."
+    },
+    {
+      "id": "recall_debug_panel",
+      "layer": "Recall and Debug",
+      "elf_evidence": "docs/evidence/benchmarking/2026-06-20-recall-debug-panel-report.md",
+      "completion_rule": "Selected, dropped, available, reviewable, not_requested, and blocked rows remain visible across memory, docs, pages, graph facts, and proposals."
+    }
+  ],
+  "product_matrix": [
+    {
+      "product": "ELF",
+      "coverage": "complete_same_repo",
+      "statuses": {
+        "source_library_ingest_hydration": "pass",
+        "memory_authority_history_read_profiles": "pass",
+        "knowledge_workspace_pages": "pass",
+        "temporal_topic_graph_lite": "pass",
+        "dreaming_review_queue": "pass",
+        "recall_debug_panel": "pass"
+      },
+      "strongest_advantage": "Policy-gated, source-linked, replayable memory and knowledge authority across all six layers.",
+      "remaining_gap": "Needs a richer operator UI and more same-corpus adapters for competitor specialties.",
+      "evidence": "README checked-in live benchmark snapshot plus June 20 component reports."
+    },
+    {
+      "product": "qmd",
+      "coverage": "partial_same_corpus",
+      "statuses": {
+        "source_library_ingest_hydration": "not_tested",
+        "memory_authority_history_read_profiles": "wrong_result",
+        "knowledge_workspace_pages": "wrong_result",
+        "temporal_topic_graph_lite": "not_tested",
+        "dreaming_review_queue": "wrong_result",
+        "recall_debug_panel": "wrong_result"
+      },
+      "strongest_advantage": "Local retrieval pipeline, query expansion, weighted fusion, rerank, short CLI replay, and top-k debug ergonomics.",
+      "remaining_gap": "Does not expose ELF-style source-authority, dropped-candidate, service trace, page rebuild, graph-lite, or review queue semantics in the current adapter.",
+      "evidence": "docs/evidence/benchmarking/2026-06-19-qmd-debug-ergonomics-dreaming-retest-report.md"
+    },
+    {
+      "product": "agentmemory",
+      "coverage": "partial_same_corpus",
+      "statuses": {
+        "source_library_ingest_hydration": "incomplete",
+        "memory_authority_history_read_profiles": "incomplete",
+        "knowledge_workspace_pages": "not_tested",
+        "temporal_topic_graph_lite": "not_tested",
+        "dreaming_review_queue": "blocked",
+        "recall_debug_panel": "not_tested"
+      },
+      "strongest_advantage": "Cross-agent hooks, MCP/REST packaging, local viewer, and coding-agent continuity workflow references.",
+      "remaining_gap": "Durable storage, lifecycle/cold-start proof, and capture breadth remain incomplete or blocked in the checked-in evidence.",
+      "evidence": "docs/evidence/benchmarking/2026-06-11-first-generation-oss-continuity-source-store-report.md"
+    },
+    {
+      "product": "OpenViking",
+      "coverage": "partial_same_corpus",
+      "statuses": {
+        "source_library_ingest_hydration": "wrong_result",
+        "memory_authority_history_read_profiles": "blocked",
+        "knowledge_workspace_pages": "not_tested",
+        "temporal_topic_graph_lite": "not_tested",
+        "dreaming_review_queue": "not_tested",
+        "recall_debug_panel": "not_tested"
+      },
+      "strongest_advantage": "Filesystem-like context URIs, hierarchy, staged trajectory, and recursive context expansion as a product idea.",
+      "remaining_gap": "Same-corpus staged trajectory, hierarchy selection, and recursive expansion stay blocked until evidence-bearing stage artifacts exist.",
+      "evidence": "docs/evidence/benchmarking/2026-06-19-openviking-trajectory-materialization-report.md"
+    },
+    {
+      "product": "mem0/OpenMemory",
+      "coverage": "partial_same_corpus",
+      "statuses": {
+        "source_library_ingest_hydration": "not_tested",
+        "memory_authority_history_read_profiles": "pass",
+        "knowledge_workspace_pages": "not_tested",
+        "temporal_topic_graph_lite": "blocked",
+        "dreaming_review_queue": "not_tested",
+        "recall_debug_panel": "blocked"
+      },
+      "strongest_advantage": "Entity-scoped memory history, hosted ecosystem, OpenMemory UI/export direction, and optional graph memory concept.",
+      "remaining_gap": "OpenMemory product UI/export remains blocked locally; hosted Platform and optional graph memory are non-goals for this lane.",
+      "evidence": "docs/evidence/benchmarking/2026-06-19-openmemory-ui-export-product-readback-report.md"
+    },
+    {
+      "product": "claude-mem",
+      "coverage": "partial_reference",
+      "statuses": {
+        "source_library_ingest_hydration": "blocked",
+        "memory_authority_history_read_profiles": "incomplete",
+        "knowledge_workspace_pages": "not_tested",
+        "temporal_topic_graph_lite": "not_tested",
+        "dreaming_review_queue": "blocked",
+        "recall_debug_panel": "not_tested"
+      },
+      "strongest_advantage": "Progressive disclosure UX, local inspection, and automatic capture-loop reference behavior.",
+      "remaining_gap": "Hook/viewer capture proof is still blocked in Docker-contained checked-in evidence.",
+      "evidence": "docs/evidence/benchmarking/2026-06-11-first-generation-oss-continuity-source-store-report.md"
+    },
+    {
+      "product": "memsearch",
+      "coverage": "partial_same_corpus",
+      "statuses": {
+        "source_library_ingest_hydration": "pass",
+        "memory_authority_history_read_profiles": "pass",
+        "knowledge_workspace_pages": "not_tested",
+        "temporal_topic_graph_lite": "not_tested",
+        "dreaming_review_queue": "not_tested",
+        "recall_debug_panel": "not_tested"
+      },
+      "strongest_advantage": "Markdown-first canonical store, incremental reindex, and practical hybrid retrieval workflow.",
+      "remaining_gap": "Broader prompt behavior, TTL, knowledge-page rebuild, graph, and review queue coverage are not encoded.",
+      "evidence": "docs/evidence/benchmarking/2026-06-11-first-generation-oss-continuity-source-store-report.md"
+    },
+    {
+      "product": "Letta",
+      "coverage": "blocked_materialization",
+      "statuses": {
+        "source_library_ingest_hydration": "not_tested",
+        "memory_authority_history_read_profiles": "blocked",
+        "knowledge_workspace_pages": "not_tested",
+        "temporal_topic_graph_lite": "not_tested",
+        "dreaming_review_queue": "blocked",
+        "recall_debug_panel": "not_tested"
+      },
+      "strongest_advantage": "Core versus archival memory model and export/readback product concept.",
+      "remaining_gap": "Contained Letta export/readback remains disabled by default and blocked until core block JSON, archival readback/search JSON, and source ids exist.",
+      "evidence": "docs/evidence/benchmarking/2026-06-19-letta-core-archive-export-readback-report.md"
+    },
+    {
+      "product": "Graphiti/Zep",
+      "coverage": "blocked_reference",
+      "statuses": {
+        "source_library_ingest_hydration": "not_tested",
+        "memory_authority_history_read_profiles": "not_tested",
+        "knowledge_workspace_pages": "not_tested",
+        "temporal_topic_graph_lite": "blocked",
+        "dreaming_review_queue": "not_tested",
+        "recall_debug_panel": "not_tested"
+      },
+      "strongest_advantage": "Temporal graph validity vocabulary and graph-memory product reference.",
+      "remaining_gap": "Representative graph/RAG slice remains blocked by provider/graph-store setup and lacks comparable same-corpus fact artifacts.",
+      "evidence": "docs/evidence/benchmarking/2026-06-19-graph-rag-citation-navigation-promotion-report.md"
+    },
+    {
+      "product": "GraphRAG",
+      "coverage": "partial_reference",
+      "statuses": {
+        "source_library_ingest_hydration": "not_tested",
+        "memory_authority_history_read_profiles": "not_tested",
+        "knowledge_workspace_pages": "incomplete",
+        "temporal_topic_graph_lite": "blocked",
+        "dreaming_review_queue": "not_tested",
+        "recall_debug_panel": "not_tested"
+      },
+      "strongest_advantage": "Graph-oriented retrieval and citation/navigation reference direction.",
+      "remaining_gap": "No contained same-corpus page/fact artifact comparable to ELF knowledge and graph reports exists yet.",
+      "evidence": "docs/evidence/benchmarking/2026-06-19-graph-rag-citation-navigation-promotion-report.md"
+    },
+    {
+      "product": "RAGFlow",
+      "coverage": "blocked_reference",
+      "statuses": {
+        "source_library_ingest_hydration": "not_tested",
+        "memory_authority_history_read_profiles": "not_tested",
+        "knowledge_workspace_pages": "blocked",
+        "temporal_topic_graph_lite": "blocked",
+        "dreaming_review_queue": "not_tested",
+        "recall_debug_panel": "not_tested"
+      },
+      "strongest_advantage": "RAG workflow and document processing product reference.",
+      "remaining_gap": "Representative graph/RAG comparison remains blocked until contained adapter outputs exist.",
+      "evidence": "docs/evidence/benchmarking/2026-06-19-graph-rag-citation-navigation-promotion-report.md"
+    },
+    {
+      "product": "LightRAG",
+      "coverage": "incomplete_reference",
+      "statuses": {
+        "source_library_ingest_hydration": "not_tested",
+        "memory_authority_history_read_profiles": "not_tested",
+        "knowledge_workspace_pages": "incomplete",
+        "temporal_topic_graph_lite": "incomplete",
+        "dreaming_review_queue": "not_tested",
+        "recall_debug_panel": "not_tested"
+      },
+      "strongest_advantage": "Lightweight graph/RAG architecture reference.",
+      "remaining_gap": "Representative comparison is incomplete, so no parity or loss claim is allowed.",
+      "evidence": "docs/evidence/benchmarking/2026-06-19-graph-rag-citation-navigation-promotion-report.md"
+    },
+    {
+      "product": "graphify",
+      "coverage": "scored_smoke",
+      "statuses": {
+        "source_library_ingest_hydration": "not_tested",
+        "memory_authority_history_read_profiles": "not_tested",
+        "knowledge_workspace_pages": "wrong_result",
+        "temporal_topic_graph_lite": "wrong_result",
+        "dreaming_review_queue": "not_tested",
+        "recall_debug_panel": "not_tested"
+      },
+      "strongest_advantage": "Tiny scored graph/RAG smoke target that can expose artifact-shape mismatches.",
+      "remaining_gap": "Current scored smoke is wrong_result; it is not graph/RAG parity evidence.",
+      "evidence": "docs/evidence/benchmarking/2026-06-19-graph-rag-citation-navigation-promotion-report.md"
+    },
+    {
+      "product": "llm-wiki",
+      "coverage": "reference_only",
+      "statuses": {
+        "source_library_ingest_hydration": "not_tested",
+        "memory_authority_history_read_profiles": "not_tested",
+        "knowledge_workspace_pages": "not_tested",
+        "temporal_topic_graph_lite": "not_tested",
+        "dreaming_review_queue": "not_tested",
+        "recall_debug_panel": "not_tested"
+      },
+      "strongest_advantage": "Compiled wiki and knowledge-page workflow reference.",
+      "remaining_gap": "No comparable same-corpus scored page artifacts are checked in.",
+      "evidence": "docs/evidence/benchmarking/2026-06-20-live-knowledge-page-rebuild-lint-report.md"
+    },
+    {
+      "product": "gbrain",
+      "coverage": "reference_only",
+      "statuses": {
+        "source_library_ingest_hydration": "not_tested",
+        "memory_authority_history_read_profiles": "not_tested",
+        "knowledge_workspace_pages": "not_tested",
+        "temporal_topic_graph_lite": "not_tested",
+        "dreaming_review_queue": "not_tested",
+        "recall_debug_panel": "not_tested"
+      },
+      "strongest_advantage": "Personal knowledge-base and query-save/lint loop reference.",
+      "remaining_gap": "No comparable scored adapter output is checked in.",
+      "evidence": "docs/evidence/benchmarking/2026-06-20-live-knowledge-page-rebuild-lint-report.md"
+    },
+    {
+      "product": "LangGraph",
+      "coverage": "blocked_reference",
+      "statuses": {
+        "source_library_ingest_hydration": "not_tested",
+        "memory_authority_history_read_profiles": "not_tested",
+        "knowledge_workspace_pages": "not_tested",
+        "temporal_topic_graph_lite": "blocked",
+        "dreaming_review_queue": "not_tested",
+        "recall_debug_panel": "not_tested"
+      },
+      "strongest_advantage": "Agent graph orchestration reference for stateful workflows.",
+      "remaining_gap": "No comparable Agent Knowledge OS memory/page/graph artifact is checked in for this matrix.",
+      "evidence": "docs/runbook/benchmarking/real_world_agent_memory_benchmark.md"
+    },
+    {
+      "product": "nanograph",
+      "coverage": "blocked_reference",
+      "statuses": {
+        "source_library_ingest_hydration": "not_tested",
+        "memory_authority_history_read_profiles": "not_tested",
+        "knowledge_workspace_pages": "not_tested",
+        "temporal_topic_graph_lite": "blocked",
+        "dreaming_review_queue": "not_tested",
+        "recall_debug_panel": "not_tested"
+      },
+      "strongest_advantage": "Typed relation and small-graph memory reference.",
+      "remaining_gap": "No comparable same-corpus typed fact artifact is checked in.",
+      "evidence": "docs/runbook/benchmarking/real_world_agent_memory_benchmark.md"
+    },
+    {
+      "product": "VectifyAI PageIndex",
+      "coverage": "reference_only",
+      "statuses": {
+        "source_library_ingest_hydration": "not_tested",
+        "memory_authority_history_read_profiles": "not_tested",
+        "knowledge_workspace_pages": "not_tested",
+        "temporal_topic_graph_lite": "not_tested",
+        "dreaming_review_queue": "not_tested",
+        "recall_debug_panel": "not_tested"
+      },
+      "strongest_advantage": "Vectorless, reasoning-based long-document tree retrieval and PageIndex MCP ecosystem direction.",
+      "remaining_gap": "No same-corpus PageIndex adapter or long-PDF tree artifact is checked into ELF yet.",
+      "evidence": "https://github.com/VectifyAI/PageIndex"
+    },
+    {
+      "product": "VectifyAI OpenKB",
+      "coverage": "reference_only",
+      "statuses": {
+        "source_library_ingest_hydration": "not_tested",
+        "memory_authority_history_read_profiles": "not_tested",
+        "knowledge_workspace_pages": "not_tested",
+        "temporal_topic_graph_lite": "not_tested",
+        "dreaming_review_queue": "not_tested",
+        "recall_debug_panel": "not_tested"
+      },
+      "strongest_advantage": "Document-to-wiki compilation, concept/entity pages, cross-links, lint reports, watch/recompile workflow, and PageIndex-backed long PDF handling.",
+      "remaining_gap": "OpenKB is a direct knowledge-library reference, but no same-corpus ELF benchmark adapter has run it yet.",
+      "evidence": "https://github.com/VectifyAI/OpenKB"
+    }
+  ],
+  "competitor_strengths": [
+    {
+      "product": "qmd",
+      "strength": "Transparent local retrieval pipeline and compact replay ergonomics.",
+      "elf_response": "Add user-facing retrieval/fusion/rerank knobs and short replay artifacts on top of the recall debug panel.",
+      "evidence_class": "wrong_result"
+    },
+    {
+      "product": "VectifyAI PageIndex",
+      "strength": "Long-document tree search without a vector database and PageIndex MCP ecosystem.",
+      "elf_response": "Add a long-document tree adapter path for Source Library and compare it against ELF source refs and page rebuilds.",
+      "evidence_class": "not_tested"
+    },
+    {
+      "product": "VectifyAI OpenKB",
+      "strength": "Compiled Markdown wiki, concept/entity pages, saved explorations, lint, watch, and recompile workflows.",
+      "elf_response": "Fold OpenKB-style library management into Knowledge Workspace without weakening ELF source-of-truth boundaries.",
+      "evidence_class": "not_tested"
+    },
+    {
+      "product": "OpenViking",
+      "strength": "Staged context trajectory, hierarchy selection, and recursive expansion.",
+      "elf_response": "Emit comparable stage artifacts from ELF recall planning and source-library hydration.",
+      "evidence_class": "blocked"
+    },
+    {
+      "product": "mem0/OpenMemory",
+      "strength": "Entity-scoped memory history, hosted ecosystem, UI/export direction, and optional graph memory.",
+      "elf_response": "Strengthen history/event APIs, export UX, and optional graph-context channel while preserving policy-gated writes.",
+      "evidence_class": "blocked"
+    },
+    {
+      "product": "Letta",
+      "strength": "Core/archive memory split and productized memory export/readback model.",
+      "elf_response": "Keep ELF core/archive source refs, but add contained adapter output before win/tie/loss claims.",
+      "evidence_class": "blocked"
+    },
+    {
+      "product": "Graphiti/Zep and graph/RAG projects",
+      "strength": "Temporal graph validity, citation/navigation, and graph retrieval product references.",
+      "elf_response": "Expand graph-lite reports into adapter-backed temporal fact comparison without replacing Postgres authority.",
+      "evidence_class": "blocked"
+    },
+    {
+      "product": "agentmemory and claude-mem",
+      "strength": "Capture hooks, local viewers, and practical continuity UX.",
+      "elf_response": "Build the operator viewer around Source Library, Memory Authority, Dreaming queue, and recall debug surfaces.",
+      "evidence_class": "incomplete"
+    }
+  ],
+  "optimization_queue": [
+    {
+      "key": "pageindex_openkb_source_library_adapter",
+      "priority": "P0",
+      "generated_from_delta": true,
+      "delta": "VectifyAI PageIndex/OpenKB are reference-only but directly target long-document library management and knowledge compilation.",
+      "next_action": "Create a contained PageIndex/OpenKB adapter over benchmark-owned sources and compare tree/wiki artifacts against ELF source refs, knowledge pages, and recall debug rows.",
+      "mapping": "issue_seed"
+    },
+    {
+      "key": "qmd_retrieval_knobs_and_short_replay",
+      "priority": "P0",
+      "generated_from_delta": true,
+      "delta": "qmd keeps the measured retrieval-debug ergonomics edge despite ELF's trace/stage visibility wins.",
+      "next_action": "Expose retrieval expansion, fusion, rerank, top-k, and compact replay artifacts in ELF recall/debug surfaces.",
+      "mapping": "docs/evidence/benchmarking/2026-06-19-qmd-debug-ergonomics-dreaming-retest-report.md"
+    },
+    {
+      "key": "operator_knowledge_library_ui",
+      "priority": "P0",
+      "generated_from_delta": true,
+      "delta": "ELF now has source, page, graph, proposal, and debug APIs but no unified knowledge-library management surface.",
+      "next_action": "Build a library UI that can ingest saved articles/threads, show source docs, derived pages, graph facts, proposal queue, and replayable recall traces.",
+      "mapping": "XY-1022 follow-up"
+    },
+    {
+      "key": "openviking_context_trajectory_artifacts",
+      "priority": "P1",
+      "generated_from_delta": true,
+      "delta": "OpenViking-style hierarchy and recursive trajectory remain blocked because comparable stage artifacts are missing.",
+      "next_action": "Emit same-corpus stage trajectory, hierarchy selection, rejected siblings, and recursive expansion artifacts.",
+      "mapping": "docs/evidence/benchmarking/2026-06-19-openviking-trajectory-materialization-report.md"
+    },
+    {
+      "key": "letta_core_archive_export_readback",
+      "priority": "P1",
+      "generated_from_delta": true,
+      "delta": "Letta core/archive comparison remains blocked until export/search/readback source ids exist.",
+      "next_action": "Run contained Letta adapter materialization with core block JSON, archival search/readback JSON, and source ids.",
+      "mapping": "docs/evidence/benchmarking/2026-06-19-letta-core-archive-export-readback-report.md"
+    },
+    {
+      "key": "openmemory_ui_export_and_history_parity",
+      "priority": "P1",
+      "generated_from_delta": true,
+      "delta": "OpenMemory UI/export remains blocked and mem0 history remains a measured reference advantage.",
+      "next_action": "Add product-container UI/export readback and strengthen ELF history/export APIs without weakening write policy.",
+      "mapping": "docs/evidence/benchmarking/2026-06-19-openmemory-ui-export-product-readback-report.md"
+    },
+    {
+      "key": "graph_rag_temporal_adapter_matrix",
+      "priority": "P1",
+      "generated_from_delta": true,
+      "delta": "Graphiti/Zep, GraphRAG, RAGFlow, LightRAG, graphify, LangGraph, and nanograph remain typed non-pass or reference-only.",
+      "next_action": "Produce contained same-corpus graph fact/page/citation artifacts while keeping ELF graph-lite as readback over source authority.",
+      "mapping": "docs/evidence/benchmarking/2026-06-19-graph-rag-citation-navigation-promotion-report.md"
+    },
+    {
+      "key": "agentmemory_claude_mem_capture_viewer",
+      "priority": "P2",
+      "generated_from_delta": true,
+      "delta": "agentmemory and claude-mem still point to practical capture/viewer UX that ELF has not productized.",
+      "next_action": "Add a local operator viewer and capture audit flow across Source Library, Memory Authority, and recall debug traces.",
+      "mapping": "docs/evidence/benchmarking/2026-06-11-first-generation-oss-continuity-source-store-report.md"
+    },
+    {
+      "key": "private_provider_production_refresh",
+      "priority": "P2",
+      "generated_from_delta": true,
+      "delta": "XY-930 proxy/public-corpus evidence cannot prove real private-corpus or provider-backed quality.",
+      "next_action": "Run operator-approved private/provider benchmark only when routed private corpus and provider setup exist.",
+      "mapping": "docs/evidence/benchmarking/2026-06-19-operator-approved-public-proxy-production-private-addendum.md"
+    }
+  ],
+  "claim_boundaries": {
+    "no_broad_superiority_claim": true,
+    "no_private_corpus_claim": true,
+    "no_hosted_provider_claim": true,
+    "reference_only_projects_do_not_count_as_pass": true,
+    "not_tested_is_not_pass": true,
+    "blocked_is_not_pass": true
+  },
+  "source_evidence": [
+    "docs/evidence/benchmarking/2026-06-20-recall-debug-panel-report.md",
+    "docs/evidence/benchmarking/2026-06-20-dreaming-review-queue-report.md",
+    "docs/evidence/benchmarking/2026-06-20-graph-topic-map-report.md",
+    "docs/evidence/benchmarking/2026-06-20-knowledge-workspace-version-diff-report.md",
+    "docs/evidence/benchmarking/2026-06-20-live-knowledge-page-rebuild-lint-report.md",
+    "docs/evidence/benchmarking/2026-06-19-qmd-debug-ergonomics-dreaming-retest-report.md",
+    "docs/evidence/benchmarking/2026-06-19-openviking-trajectory-materialization-report.md",
+    "docs/evidence/benchmarking/2026-06-19-letta-core-archive-export-readback-report.md",
+    "docs/evidence/benchmarking/2026-06-19-openmemory-ui-export-product-readback-report.md",
+    "docs/evidence/benchmarking/2026-06-19-graph-rag-citation-navigation-promotion-report.md",
+    "https://github.com/VectifyAI/PageIndex",
+    "https://github.com/VectifyAI/OpenKB",
+    "https://github.com/VectifyAI"
+  ],
+  "validation_expectations": [
+    "cargo make fmt",
+    "cargo make test",
+    "cargo make check-docs",
+    "decodex docs check",
+    "cargo test -p elf-eval --test real_world_job_benchmark",
+    "git diff --check"
+  ]
+}
diff --git a/apps/elf-eval/tests/real_world_job_benchmark.rs b/apps/elf-eval/tests/real_world_job_benchmark.rs
index e17c6bc3..d57f838d 100644
--- a/apps/elf-eval/tests/real_world_job_benchmark.rs
+++ b/apps/elf-eval/tests/real_world_job_benchmark.rs
@@ -258,6 +258,10 @@ fn recall_debug_panel_report_json_path() -> Result<PathBuf> {
 	report_snapshot_path("2026-06-20-recall-debug-panel-report.json")
 }
 
+fn agent_knowledge_os_closeout_benchmark_report_json_path() -> Result<PathBuf> {
+	report_snapshot_path("2026-06-20-agent-knowledge-os-closeout-benchmark-report.json")
+}
+
 fn openmemory_ui_export_product_readback_report_json_path() -> Result<PathBuf> {
 	report_snapshot_path("2026-06-19-openmemory-ui-export-product-readback-report.json")
 }
@@ -312,6 +316,14 @@ fn recall_debug_panel_report_markdown_path() -> Result<PathBuf> {
 		.join("2026-06-20-recall-debug-panel-report.md"))
 }
 
+fn agent_knowledge_os_closeout_benchmark_report_markdown_path() -> Result<PathBuf> {
+	Ok(workspace_root()?
+		.join("docs")
+		.join("evidence")
+		.join("benchmarking")
+		.join("2026-06-20-agent-knowledge-os-closeout-benchmark-report.md"))
+}
+
 fn openmemory_ui_export_product_readback_report_markdown_path() -> Result<PathBuf> {
 	Ok(workspace_root()?
 		.join("docs")
@@ -3809,6 +3821,165 @@ fn recall_debug_panel_report_wires_cross_layer_debug_contract() -> Result<()> {
 	Ok(())
 }
 
+#[test]
+fn agent_knowledge_os_closeout_benchmark_preserves_full_matrix_boundaries() -> Result<()> {
+	let report = serde_json::from_str::<Value>(&fs::read_to_string(
+		agent_knowledge_os_closeout_benchmark_report_json_path()?,
+	)?)?;
+
+	assert_eq!(
+		report.pointer("/schema").and_then(Value::as_str),
+		Some("elf.agent_knowledge_os_closeout_benchmark_report/v1")
+	);
+	assert_eq!(report.pointer("/authority").and_then(Value::as_str), Some("XY-1023"));
+	assert_eq!(
+		report.pointer("/summary/strongest_measured_integrated_product").and_then(Value::as_str),
+		Some("ELF integrated Agent Knowledge OS")
+	);
+	assert_eq!(
+		report.pointer("/all_project_fixture_rerun/status").and_then(Value::as_str),
+		Some("pass")
+	);
+	assert_eq!(
+		report.pointer("/all_project_fixture_rerun/job_count").and_then(Value::as_u64),
+		Some(62)
+	);
+	assert_eq!(report.pointer("/all_project_fixture_rerun/pass").and_then(Value::as_u64), Some(55));
+	assert_eq!(report.pointer("/summary/product_count").and_then(Value::as_u64), Some(19));
+	assert_eq!(report.pointer("/summary/scenario_count").and_then(Value::as_u64), Some(6));
+	assert_eq!(
+		report
+			.pointer("/summary/not_every_product_has_complete_live_coverage")
+			.and_then(Value::as_bool),
+		Some(true)
+	);
+	assert_eq!(
+		report.pointer("/summary/evidence_class_counts/pass").and_then(Value::as_u64),
+		Some(9)
+	);
+	assert_eq!(
+		report.pointer("/summary/evidence_class_counts/not_tested").and_then(Value::as_u64),
+		Some(78)
+	);
+
+	let scenarios = array_at(&report, "/supported_scenarios")?;
+	let matrix = array_at(&report, "/product_matrix")?;
+
+	for scenario in [
+		"source_library_ingest_hydration",
+		"memory_authority_history_read_profiles",
+		"knowledge_workspace_pages",
+		"temporal_topic_graph_lite",
+		"dreaming_review_queue",
+		"recall_debug_panel",
+	] {
+		find_by_field(scenarios, "/id", scenario)?;
+	}
+
+	let elf = find_by_field(matrix, "/product", "ELF")?;
+
+	for scenario in [
+		"source_library_ingest_hydration",
+		"memory_authority_history_read_profiles",
+		"knowledge_workspace_pages",
+		"temporal_topic_graph_lite",
+		"dreaming_review_queue",
+		"recall_debug_panel",
+	] {
+		assert_eq!(
+			elf.pointer(&format!("/statuses/{scenario}")).and_then(Value::as_str),
+			Some("pass")
+		);
+	}
+
+	let qmd = find_by_field(matrix, "/product", "qmd")?;
+
+	assert_eq!(
+		qmd.pointer("/statuses/recall_debug_panel").and_then(Value::as_str),
+		Some("wrong_result")
+	);
+	assert!(
+		qmd.pointer("/strongest_advantage")
+			.and_then(Value::as_str)
+			.is_some_and(|value| value.contains("weighted fusion"))
+	);
+
+	for product in ["VectifyAI PageIndex", "VectifyAI OpenKB"] {
+		let row = find_by_field(matrix, "/product", product)?;
+
+		assert_eq!(row.pointer("/coverage").and_then(Value::as_str), Some("reference_only"));
+		assert_eq!(
+			row.pointer("/statuses/knowledge_workspace_pages").and_then(Value::as_str),
+			Some("not_tested")
+		);
+	}
+
+	assert_eq!(
+		report.pointer("/claim_boundaries/no_broad_superiority_claim").and_then(Value::as_bool),
+		Some(true)
+	);
+	assert_eq!(
+		report
+			.pointer("/claim_boundaries/reference_only_projects_do_not_count_as_pass")
+			.and_then(Value::as_bool),
+		Some(true)
+	);
+	assert!(array_contains_str(
+		&report,
+		"/source_evidence",
+		"https://github.com/VectifyAI/PageIndex"
+	)?);
+	assert!(array_contains_str(
+		&report,
+		"/source_evidence",
+		"https://github.com/VectifyAI/OpenKB"
+	)?);
+
+	Ok(())
+}
+
+#[test]
+fn agent_knowledge_os_closeout_benchmark_wires_docs_and_optimization_queue() -> Result<()> {
+	let report = serde_json::from_str::<Value>(&fs::read_to_string(
+		agent_knowledge_os_closeout_benchmark_report_json_path()?,
+	)?)?;
+	let markdown =
+		fs::read_to_string(agent_knowledge_os_closeout_benchmark_report_markdown_path()?)?;
+	let benchmarking_index = fs::read_to_string(benchmarking_index_path()?)?;
+	let readme = fs::read_to_string(readme_path()?)?;
+	let queue = array_at(&report, "/optimization_queue")?;
+
+	for item in queue {
+		assert_eq!(item.pointer("/generated_from_delta").and_then(Value::as_bool), Some(true));
+	}
+	for key in [
+		"pageindex_openkb_source_library_adapter",
+		"qmd_retrieval_knobs_and_short_replay",
+		"operator_knowledge_library_ui",
+		"openviking_context_trajectory_artifacts",
+		"graph_rag_temporal_adapter_matrix",
+	] {
+		let item = find_by_field(queue, "/key", key)?;
+
+		assert_eq!(item.pointer("/generated_from_delta").and_then(Value::as_bool), Some(true));
+	}
+
+	assert!(markdown.contains("ELF is the strongest measured integrated product"));
+	assert!(markdown.contains("complete live coverage"));
+	assert!(markdown.contains("VectifyAI PageIndex"));
+	assert!(markdown.contains("VectifyAI OpenKB"));
+	assert!(markdown.contains("Do not claim ELF broadly beats every competitor"));
+	assert!(
+		benchmarking_index.contains("2026-06-20-agent-knowledge-os-closeout-benchmark-report.md")
+	);
+	assert!(readme.contains("Agent Knowledge OS closeout after XY-1023"));
+	assert!(readme.contains("62 jobs, 55 pass"));
+	assert!(readme.contains("VectifyAI PageIndex/OpenKB"));
+	assert!(readme.contains("strongest measured integrated"));
+
+	Ok(())
+}
+
 #[test]
 fn operator_approved_public_proxy_private_addendum_preserves_boundary() -> Result<()> {
 	let report = serde_json::from_str::<Value>(&fs::read_to_string(
diff --git a/docs/evidence/benchmarking/2026-06-20-agent-knowledge-os-closeout-benchmark-report.md b/docs/evidence/benchmarking/2026-06-20-agent-knowledge-os-closeout-benchmark-report.md
new file mode 100644
index 00000000..68136d15
--- /dev/null
+++ b/docs/evidence/benchmarking/2026-06-20-agent-knowledge-os-closeout-benchmark-report.md
@@ -0,0 +1,148 @@
+---
+type: Evidence
+title: "Agent Knowledge OS Closeout Benchmark Report - June 20, 2026"
+description: "Checked-in closeout evidence matrix for the Agent Knowledge OS program."
+resource: docs/evidence/benchmarking/2026-06-20-agent-knowledge-os-closeout-benchmark-report.md
+status: active
+authority: current_state
+owner: evidence
+last_verified: 2026-06-20
+tags:
+  - docs
+  - evidence
+  - benchmarking
+  - agent-knowledge-os
+---
+# Agent Knowledge OS Closeout Benchmark Report - June 20, 2026
+
+Goal: Close XY-1023 by publishing a table-driven self-assessment after the staged
+Agent Knowledge OS lanes landed.
+
+Inputs: `apps/elf-eval/fixtures/report_snapshots/2026-06-20-agent-knowledge-os-closeout-benchmark-report.json`,
+the June 20 component reports, the June 19 competitor retests/materialization
+reports, and the current VectifyAI PageIndex/OpenKB GitHub readback.
+
+## Command Evidence
+
+| Command | Status | Result |
+| --- | --- | --- |
+| `cargo make real-world-memory` | pass | Reran the checked-in all-project fixture suite: 62 jobs, 17 encoded suites, 55 pass, 0 wrong_result, 0 incomplete, 7 blocked, 1.000 evidence/source-ref/quote coverage, mean score 0.887. |
+| `cargo test -p elf-eval --test real_world_job_benchmark agent_knowledge_os_closeout_benchmark -- --nocapture` | pass | Guards the XY-1023 summary counts, key matrix boundaries, VectifyAI reference-only rows, claim boundaries, README/index links, and optimization queue. |
+
+## Executive Judgment
+
+ELF is the strongest measured integrated product in the current checked-in Agent
+Knowledge OS matrix. It is the only product with same-repo evidence across all six
+layers: Source Library, Memory Authority, Knowledge Workspace, graph-lite facts,
+Dreaming review queue, and recall/debug panel.
+
+That is not a broad "ELF beats everyone everywhere" claim. Not every product has
+complete live coverage. qmd remains the retrieval/debug ergonomics reference;
+OpenViking remains the context-trajectory reference; mem0/OpenMemory remains the
+entity-history and ecosystem reference; Letta remains the core/archive memory
+reference; Graphiti/Zep and graph/RAG projects remain graph-memory references; and
+VectifyAI PageIndex/OpenKB are now explicit reference-only competitors for long
+document tree retrieval and knowledge-base compilation.
+
+## Coverage Summary
+
+| Metric | Value |
+| --- | --- |
+| Products/projects in matrix | 19 |
+| Agent Knowledge OS scenarios | 6 |
+| Complete same-repo product coverage | ELF only |
+| Matrix pass cells | 9 |
+| Matrix wrong_result cells | 7 |
+| Matrix incomplete cells | 6 |
+| Matrix blocked cells | 14 |
+| Matrix not_tested cells | 78 |
+
+Evidence classes keep their normal meaning: `pass` means checked-in evidence
+supports the claim; `wrong_result` means the adapter ran but missed required
+evidence; `incomplete` means partial behavior or artifact coverage; `blocked` means
+the required setup/artifact is missing; `not_tested` means reference-only or no
+same-corpus benchmark coverage.
+
+## Product Scenario Matrix
+
+| Product/project | Coverage | Source Library | Memory Authority | Knowledge Workspace | Graph-lite/Temporal | Dreaming Review | Recall Debug | Current strongest advantage |
+| --- | --- | --- | --- | --- | --- | --- | --- | --- |
+| ELF | complete_same_repo | pass | pass | pass | pass | pass | pass | Policy-gated, source-linked, replayable authority across all six layers. |
+| qmd | partial_same_corpus | not_tested | wrong_result | wrong_result | not_tested | wrong_result | wrong_result | Query expansion, weighted fusion, rerank, compact replay, and local debug knobs. |
+| agentmemory | partial_same_corpus | incomplete | incomplete | not_tested | not_tested | blocked | not_tested | Cross-agent hooks, MCP/REST packaging, local viewer, continuity workflow. |
+| OpenViking | partial_same_corpus | wrong_result | blocked | not_tested | not_tested | not_tested | not_tested | Filesystem-like context URIs, hierarchy, staged retrieval trajectory. |
+| mem0/OpenMemory | partial_same_corpus | not_tested | pass | not_tested | blocked | not_tested | blocked | Entity-scoped memory history, hosted ecosystem, OpenMemory UI/export direction. |
+| claude-mem | partial_reference | blocked | incomplete | not_tested | not_tested | blocked | not_tested | Progressive disclosure UX, local viewer, automatic capture-loop reference. |
+| memsearch | partial_same_corpus | pass | pass | not_tested | not_tested | not_tested | not_tested | Markdown-first canonical store, incremental reindex, hybrid retrieval. |
+| Letta | blocked_materialization | not_tested | blocked | not_tested | not_tested | blocked | not_tested | Core/archive memory model and export/readback product concept. |
+| Graphiti/Zep | blocked_reference | not_tested | not_tested | not_tested | blocked | not_tested | not_tested | Temporal graph validity vocabulary and graph-memory product reference. |
+| GraphRAG | partial_reference | not_tested | not_tested | incomplete | blocked | not_tested | not_tested | Graph-oriented retrieval and citation/navigation reference direction. |
+| RAGFlow | blocked_reference | not_tested | not_tested | blocked | blocked | not_tested | not_tested | RAG workflow and document processing product reference. |
+| LightRAG | incomplete_reference | not_tested | not_tested | incomplete | incomplete | not_tested | not_tested | Lightweight graph/RAG architecture reference. |
+| graphify | scored_smoke | not_tested | not_tested | wrong_result | wrong_result | not_tested | not_tested | Tiny scored graph/RAG smoke target for artifact-shape mismatches. |
+| llm-wiki | reference_only | not_tested | not_tested | not_tested | not_tested | not_tested | not_tested | Compiled wiki and knowledge-page workflow reference. |
+| gbrain | reference_only | not_tested | not_tested | not_tested | not_tested | not_tested | not_tested | Personal knowledge-base and query-save/lint loop reference. |
+| LangGraph | blocked_reference | not_tested | not_tested | not_tested | blocked | not_tested | not_tested | Agent graph orchestration reference for stateful workflows. |
+| nanograph | blocked_reference | not_tested | not_tested | not_tested | blocked | not_tested | not_tested | Typed relation and small-graph memory reference. |
+| VectifyAI PageIndex | reference_only | not_tested | not_tested | not_tested | not_tested | not_tested | not_tested | Vectorless long-document tree retrieval and PageIndex MCP ecosystem direction. |
+| VectifyAI OpenKB | reference_only | not_tested | not_tested | not_tested | not_tested | not_tested | not_tested | Document-to-wiki compilation, concept/entity pages, lint, watch, and recompile workflow. |
+
+## Competitor Strengths To Preserve
+
+| Competitor | Strength | Evidence class | ELF response |
+| --- | --- | --- | --- |
+| qmd | Transparent local retrieval pipeline and compact replay ergonomics. | wrong_result | Add retrieval expansion, fusion, rerank, top-k, and compact replay controls on top of recall/debug. |
+| VectifyAI PageIndex | Long-document tree search without a vector database and PageIndex MCP ecosystem. | not_tested | Add a benchmark-owned long-document tree adapter and compare it with ELF source refs and page rebuilds. |
+| VectifyAI OpenKB | Compiled Markdown wiki, concept/entity pages, saved explorations, lint, watch, and recompile workflows. | not_tested | Fold OpenKB-style library management into Knowledge Workspace without weakening source-of-truth boundaries. |
+| OpenViking | Staged context trajectory, hierarchy selection, and recursive expansion. | blocked | Emit comparable recall-planning stage artifacts, rejected siblings, and recursive expansion evidence. |
+| mem0/OpenMemory | Entity-scoped memory history, hosted ecosystem, UI/export, and optional graph memory direction. | blocked | Strengthen history/event APIs, export UX, and optional graph-context channel while preserving policy-gated writes. |
+| Letta | Core/archive memory split and memory export/readback product model. | blocked | Keep ELF core/archive source refs, but add contained adapter output before win/tie/loss claims. |
+| Graphiti/Zep and graph/RAG projects | Temporal graph validity, citation/navigation, and graph retrieval references. | blocked | Expand graph-lite reports into adapter-backed temporal fact comparison without replacing Postgres authority. |
+| agentmemory and claude-mem | Capture hooks, local viewers, and practical continuity UX. | incomplete | Build the operator viewer around Source Library, Memory Authority, Dreaming queue, and recall debug surfaces. |
+
+## ELF Advantages
+
+ELF's current durable advantage is composition: the layers are independently typed,
+source-linked, and replayable, but they now fit together as an Agent Knowledge OS.
+Source Library records preserve captured material, Memory Authority controls what
+becomes memory, Knowledge Workspace pages stay derived and linted, graph-lite facts
+stay source-backed, Dreaming proposals stay reviewable, and the recall/debug panel
+shows selected, dropped, available, reviewable, not_requested, and blocked context.
+
+This combination is stronger than any single measured competitor in the current
+matrix. It is also more conservative: ELF refuses to call reference-only strengths a
+pass, and it keeps private/provider production proof separate from simulated or
+public-proxy evidence.
+
+## Optimization Queue
+
+| Priority | Queue item | Generated from benchmark delta | Next action |
+| --- | --- | --- | --- |
+| P0 | `pageindex_openkb_source_library_adapter` | PageIndex/OpenKB are reference-only but directly target long-document library management and knowledge compilation. | Create a contained adapter over benchmark-owned sources and compare tree/wiki artifacts against ELF source refs, knowledge pages, and recall debug rows. |
+| P0 | `qmd_retrieval_knobs_and_short_replay` | qmd keeps the measured retrieval-debug ergonomics edge. | Expose retrieval expansion, fusion, rerank, top-k, and compact replay artifacts in ELF recall/debug surfaces. |
+| P0 | `operator_knowledge_library_ui` | ELF has APIs but no unified library management surface. | Build a UI for saved articles/threads, source docs, derived pages, graph facts, proposal queue, and replayable recall traces. |
+| P1 | `openviking_context_trajectory_artifacts` | OpenViking trajectory/hierarchy/recursive expansion remains blocked. | Emit same-corpus stage trajectory, hierarchy selection, rejected siblings, and recursive expansion artifacts. |
+| P1 | `letta_core_archive_export_readback` | Letta core/archive comparison remains blocked. | Run contained Letta export/readback with core block JSON, archival search/readback JSON, and source ids. |
+| P1 | `openmemory_ui_export_and_history_parity` | OpenMemory UI/export remains blocked and mem0 history remains a reference advantage. | Add product-container UI/export readback and strengthen ELF history/export APIs. |
+| P1 | `graph_rag_temporal_adapter_matrix` | Graph/RAG projects remain typed non-pass or reference-only. | Produce contained same-corpus graph fact/page/citation artifacts while keeping ELF graph-lite source-backed. |
+| P2 | `agentmemory_claude_mem_capture_viewer` | Capture/viewer UX remains incomplete or blocked. | Add a local operator viewer and capture audit flow across Source Library, Memory Authority, and recall debug traces. |
+| P2 | `private_provider_production_refresh` | XY-930 proxy/public-corpus evidence cannot prove real private-corpus or provider-backed quality. | Run only when routed private corpus and provider setup exist. |
+
+## Claim Boundaries
+
+Allowed:
+
+- ELF is the strongest measured integrated Agent Knowledge OS product in this
+  checked-in matrix.
+- ELF has complete same-repo evidence across the six Agent Knowledge OS layers.
+- qmd, PageIndex/OpenKB, OpenViking, mem0/OpenMemory, Letta, graph/RAG systems, and
+  capture/viewer projects still provide important optimization direction.
+
+Not allowed:
+
+- Do not claim ELF broadly beats every competitor on every competitor-owned strength.
+- Do not treat `not_tested`, `blocked`, `incomplete`, or `wrong_result` as pass.
+- Do not count VectifyAI PageIndex or OpenKB as benchmark wins until a same-corpus
+  adapter emits checked-in artifacts.
+- Do not claim private-corpus or hosted-provider production quality from public-proxy
+  or local fixture evidence.
diff --git a/docs/evidence/benchmarking/index.md b/docs/evidence/benchmarking/index.md
index 6500a2e9..4888ca21 100644
--- a/docs/evidence/benchmarking/index.md
+++ b/docs/evidence/benchmarking/index.md
@@ -48,3 +48,4 @@ Routes to: Benchmarking evidence concepts under `docs/evidence/benchmarking/`.
 - `2026-06-20-knowledge-workspace-version-diff-report.md`: Knowledge Workspace Version-Diff Report - June 20, 2026; proves ELF knowledge pages now expose previous-version diff metadata without perturbing page content hashes while preserving citation, lint, and source-of-truth boundaries.
 - `2026-06-20-live-knowledge-page-rebuild-lint-report.md`: Live Knowledge-Page Rebuild/Lint Report - June 20, 2026; adds a Docker-contained ELF service-native knowledge-page materialization command while preserving llm-wiki, gbrain, GraphRAG, RAGFlow, LightRAG, and graphify as separate comparison targets until they emit comparable scored page artifacts.
 - `2026-06-20-recall-debug-panel-report.md`: Recall Debug Panel Report - June 20, 2026; adds `elf.recall_debug_panel/v1` as a typed cross-layer readback over memory traces, Source Library document candidates, Knowledge Workspace pages, graph facts, and Dreaming proposals while preserving not-requested and non-pass evidence classes.
+- `2026-06-20-agent-knowledge-os-closeout-benchmark-report.md`: Agent Knowledge OS Closeout Benchmark Report - June 20, 2026; publishes the XY-1023 full product/scenario matrix, names ELF as the strongest measured integrated product, preserves qmd/OpenViking/mem0/OpenMemory/Letta/graph-RAG/VectifyAI strengths, and turns material non-pass or reference-only deltas into optimization queue items.