diff --git a/Makefile.toml b/Makefile.toml
index 5c89f94d..6e8e6c56 100644
--- a/Makefile.toml
+++ b/Makefile.toml
@@ -418,6 +418,7 @@ args = [
 # | real-world-memory-consolidation        | composite |     |
 # | real-world-memory-consolidation-json   | command   |     |
 # | real-world-memory-consolidation-report | command   |     |
+# | real-world-memory-live-consolidation   | command   |     |
 # | real-world-job-operator-ux             | composite |     |
 # | real-world-job-operator-ux-json        | command   |     |
 # | real-world-job-operator-ux-report      | command   |     |
@@ -830,6 +831,14 @@ args = [
 	"tmp/real-world-memory/consolidation/report.md",
 ]
 
+[tasks.real-world-memory-live-consolidation]
+workspace = false
+command = "bash"
+args = [
+	"-lc",
+	"docker compose -f docker-compose.baseline.yml run --build --rm -e ELF_CONSOLIDATION_LIVE_REPORT_DIR -e ELF_CONSOLIDATION_LIVE_FIXTURES baseline-runner bash scripts/real-world-consolidation-live-adapter.sh",
+]
+
 [tasks.real-world-memory-core-archival]
 workspace = false
 dependencies = [
diff --git a/README.md b/README.md
index a4cae687..aa3b0350 100644
--- a/README.md
+++ b/README.md
@@ -181,6 +181,14 @@ provider-backed ELF evidence was required.
   evidence fields. qmd remains `wrong_result` on the same slice, but this is not a
   broad qmd, Graphiti/Zep, mem0/OpenMemory, Letta, hosted-memory, or private-corpus
   superiority claim.
+- Live consolidation proposal scoring after XY-934: `cargo make
+  real-world-memory-live-consolidation` runs the consolidation fixture slice through
+  `ElfService` consolidation run creation, worker proposal materialization, and
+  apply/defer/discard review audit transitions. ELF passes 4/4 live consolidation jobs
+  with complete lineage, one unsupported-claim flag preserved, and zero source
+  mutations. Managed dreaming and Always-On Memory Agent patterns remain product
+  references, not direct live competitors, because no contained runner emits comparable
+  artifacts.
 - Live operator-debugging slice after XY-932: `cargo make
   real-world-job-operator-ux-live-adapters` emits narrow Docker-isolated
   `live_real_world` records for ELF and qmd over the operator-debugging fixtures.
@@ -255,6 +263,7 @@ Detailed evidence and interpretation:
 - [Graph/RAG Scored Smoke Adapter Report - June 11, 2026](docs/guide/benchmarking/2026-06-11-graph-rag-scored-smoke-adapter-report.md)
 - [mem0/OpenMemory History and UI Export Report - June 11, 2026](docs/guide/benchmarking/2026-06-11-mem0-openmemory-history-ui-export-report.md)
 - [Capture/Write-Policy Live Report - June 11, 2026](docs/guide/benchmarking/2026-06-11-capture-write-policy-live-report.md)
+- [Live Consolidation Proposal Scoring Report - June 16, 2026](docs/guide/benchmarking/2026-06-16-live-consolidation-proposal-scoring-report.md)
 - [First-Generation OSS Continuity and Source-Store Report - June 11, 2026](docs/guide/benchmarking/2026-06-11-first-generation-oss-continuity-source-store-report.md)
 - [Live Temporal Reconciliation Report - June 16, 2026](docs/guide/benchmarking/2026-06-16-live-temporal-reconciliation-report.md)
 - [Live Baseline Benchmark Runbook](docs/guide/benchmarking/live_baseline_benchmark.md)
@@ -335,6 +344,7 @@ Detailed comparison, mechanism-level analysis, and source map:
 - [Graph/RAG Scored Smoke Adapter Report - June 11, 2026](docs/guide/benchmarking/2026-06-11-graph-rag-scored-smoke-adapter-report.md)
 - [mem0/OpenMemory History and UI Export Report - June 11, 2026](docs/guide/benchmarking/2026-06-11-mem0-openmemory-history-ui-export-report.md)
 - [Capture/Write-Policy Live Report - June 11, 2026](docs/guide/benchmarking/2026-06-11-capture-write-policy-live-report.md)
+- [Live Consolidation Proposal Scoring Report - June 16, 2026](docs/guide/benchmarking/2026-06-16-live-consolidation-proposal-scoring-report.md)
 - [First-Generation OSS Continuity and Source-Store Report - June 11, 2026](docs/guide/benchmarking/2026-06-11-first-generation-oss-continuity-source-store-report.md)
 - [Live Temporal Reconciliation Report - June 16, 2026](docs/guide/benchmarking/2026-06-16-live-temporal-reconciliation-report.md)
 - [Live Baseline Benchmark Runbook](docs/guide/benchmarking/live_baseline_benchmark.md)
@@ -347,7 +357,8 @@ Detailed comparison, mechanism-level analysis, and source map:
 - [RAG/Graph Adapter Feasibility Research Run](docs/research/2026-06-10-xy-882-rag-graph-adapter-feasibility.json)
 
 Latest real-world benchmark report: June 16, 2026. Latest external research refresh:
-June 11, 2026.
+June 11, 2026; June 16 adds live temporal reconciliation and live consolidation
+self-check evidence.
 
 ## Documentation
 
diff --git a/apps/elf-eval/tests/real_world_job_benchmark.rs b/apps/elf-eval/tests/real_world_job_benchmark.rs
index 9ff7a7f7..a7ea546b 100644
--- a/apps/elf-eval/tests/real_world_job_benchmark.rs
+++ b/apps/elf-eval/tests/real_world_job_benchmark.rs
@@ -175,6 +175,21 @@ fn capture_write_policy_live_markdown_path() -> Result<PathBuf> {
 		.join("2026-06-11-capture-write-policy-live-report.md"))
 }
 
+fn live_consolidation_proposal_scoring_report_path() -> Result<PathBuf> {
+	Ok(workspace_root()?
+		.join("docs")
+		.join("research")
+		.join("2026-06-16-live-consolidation-proposal-scoring-report.json"))
+}
+
+fn live_consolidation_proposal_scoring_markdown_path() -> Result<PathBuf> {
+	Ok(workspace_root()?
+		.join("docs")
+		.join("guide")
+		.join("benchmarking")
+		.join("2026-06-16-live-consolidation-proposal-scoring-report.md"))
+}
+
 fn temporal_history_competitor_gap_json_path() -> Result<PathBuf> {
 	Ok(workspace_root()?
 		.join("docs")
@@ -2021,6 +2036,124 @@ fn capture_write_policy_live_report_preserves_competitor_boundaries() -> Result<
 	Ok(())
 }
 
+#[test]
+fn live_consolidation_report_preserves_reviewable_output_boundaries() -> Result<()> {
+	let workspace = workspace_root()?;
+	let report = serde_json::from_str::<Value>(&fs::read_to_string(
+		live_consolidation_proposal_scoring_report_path()?,
+	)?)?;
+	let markdown = fs::read_to_string(live_consolidation_proposal_scoring_markdown_path()?)?;
+	let benchmarking_index = fs::read_to_string(benchmarking_index_path()?)?;
+	let readme = fs::read_to_string(readme_path()?)?;
+	let benchmark_guide = fs::read_to_string(
+		workspace
+			.join("docs")
+			.join("guide")
+			.join("benchmarking")
+			.join("real_world_agent_memory_benchmark.md"),
+	)?;
+	let makefile = fs::read_to_string(workspace.join("Makefile.toml"))?;
+	let live_script =
+		fs::read_to_string(workspace.join("scripts/real-world-consolidation-live-adapter.sh"))?;
+	let live_adapter =
+		fs::read_to_string(workspace.join("apps/elf-eval/src/bin/real_world_live_adapter.rs"))?;
+
+	assert_eq!(
+		report.pointer("/schema").and_then(Value::as_str),
+		Some("elf.live_consolidation_proposal_scoring_report/v1")
+	);
+	assert_eq!(report.pointer("/authority").and_then(Value::as_str), Some("XY-934"));
+	assert_eq!(
+		report
+			.pointer("/live_consolidation_results/elf_live_real_world/suite_status")
+			.and_then(Value::as_str),
+		Some("pass")
+	);
+	assert_eq!(
+		report
+			.pointer("/live_consolidation_results/elf_live_real_world/encoded_job_count")
+			.and_then(Value::as_u64),
+		Some(4)
+	);
+	assert_eq!(
+		report
+			.pointer("/live_consolidation_results/elf_live_real_world/proposal_count")
+			.and_then(Value::as_u64),
+		Some(4)
+	);
+	assert_eq!(
+		report
+			.pointer("/live_consolidation_results/elf_live_real_world/source_mutation_count")
+			.and_then(Value::as_u64),
+		Some(0)
+	);
+	assert_eq!(
+		report
+			.pointer("/live_consolidation_results/elf_live_real_world/review_event_count")
+			.and_then(Value::as_u64),
+		Some(6)
+	);
+	assert_eq!(
+		report
+			.pointer("/live_consolidation_results/qmd_live_real_world/suite_status")
+			.and_then(Value::as_str),
+		Some("not_encoded")
+	);
+
+	let jobs = array_at(&report, "/jobs")?;
+	let project_summary =
+		find_by_field(jobs, "/job_id", "consolidation-project-summary-apply-001")?;
+	let preference =
+		find_by_field(jobs, "/job_id", "consolidation-preference-candidate-defer-001")?;
+	let contradiction =
+		find_by_field(jobs, "/job_id", "consolidation-contradiction-report-discard-001")?;
+
+	assert_eq!(
+		project_summary.pointer("/final_review_state").and_then(Value::as_str),
+		Some("applied")
+	);
+	assert_eq!(project_summary.pointer("/review_event_count").and_then(Value::as_u64), Some(2));
+	assert_eq!(preference.pointer("/final_review_state").and_then(Value::as_str), Some("archived"));
+	assert_eq!(
+		contradiction.pointer("/final_review_state").and_then(Value::as_str),
+		Some("rejected")
+	);
+	assert_eq!(
+		contradiction.pointer("/unsupported_claim_flag_count").and_then(Value::as_u64),
+		Some(1)
+	);
+	assert_eq!(contradiction.pointer("/source_lineage_count").and_then(Value::as_u64), Some(3));
+
+	let positions = array_at(&report, "/reference_positions")?;
+	let qmd = find_by_field(positions, "/project", "qmd")?;
+	let managed = find_by_field(positions, "/project", "managed_dreaming_memory_systems")?;
+	let always_on = find_by_field(positions, "/project", "always_on_memory_agent_patterns")?;
+
+	assert_eq!(qmd.pointer("/position").and_then(Value::as_str), Some("untested"));
+	assert_eq!(managed.pointer("/position").and_then(Value::as_str), Some("product_reference"));
+	assert_eq!(always_on.pointer("/position").and_then(Value::as_str), Some("product_reference"));
+	assert!(markdown.contains("ELF now has service-backed live consolidation proposal scoring"));
+	assert!(markdown.contains("This is not scheduled production consolidation"));
+	assert!(markdown.contains("Source mutations"));
+	assert!(markdown.contains("Do not mix knowledge-page rebuild/lint scoring"));
+	assert!(
+		benchmarking_index.contains("2026-06-16-live-consolidation-proposal-scoring-report.md")
+	);
+	assert!(readme.contains("Live Consolidation Proposal Scoring Report - June 16, 2026"));
+	assert!(readme.contains("real-world-memory-live-consolidation"));
+	assert!(benchmark_guide.contains("Current live consolidation increment"));
+	assert!(benchmark_guide.contains("tmp/real-world-memory/live-consolidation/summary.json"));
+	assert!(makefile.contains("[tasks.real-world-memory-live-consolidation]"));
+	assert!(makefile.contains("scripts/real-world-consolidation-live-adapter.sh"));
+	assert!(live_script.contains("elf.real_world_consolidation_live_adapter_sweep/v1"));
+	assert!(live_script.contains("real_world_live_adapter -- elf"));
+	assert!(!live_script.contains("real_world_live_adapter -- qmd"));
+	assert!(live_adapter.contains("fn materialize_elf_consolidation("));
+	assert!(live_adapter.contains("ConsolidationProposalReviewRequest"));
+
+	Ok(())
+}
+
 fn assert_live_sweep_record(adapter: &Value, production_ops_status: &str) -> Result<()> {
 	let suites = array_at(adapter, "/suites")?;
 	let capabilities = array_at(adapter, "/capabilities")?;
@@ -3016,6 +3149,7 @@ fn assert_competitor_strength_matrix_scenario_json(scenarios: &[Value]) -> Resul
 	let work_resume = find_by_field(scenarios, "/scenario_id", "work_resume")?;
 	let operator_debug = find_by_field(scenarios, "/scenario_id", "operator_debugging")?;
 	let context_trajectory = find_by_field(scenarios, "/scenario_id", "context_trajectory")?;
+	let consolidation = find_by_field(scenarios, "/scenario_id", "consolidation")?;
 
 	assert!(
 		retrieval_debug
@@ -3051,6 +3185,20 @@ fn assert_competitor_strength_matrix_scenario_json(scenarios: &[Value]) -> Resul
 			.and_then(Value::as_str)
 			.is_some_and(|claim| claim.contains("OpenMemory and claude-mem UI/export"))
 	);
+	assert!(
+		consolidation
+			.pointer("/current_elf_evidence")
+			.and_then(Value::as_str)
+			.is_some_and(|claim| claim.contains("XY-934 adds live_real_world")
+				&& claim.contains("zero source mutations"))
+	);
+	assert!(
+		consolidation
+			.pointer("/current_competitor_evidence")
+			.and_then(Value::as_str)
+			.is_some_and(|claim| claim.contains("qmd remains not_encoded")
+				&& claim.contains("product references only"))
+	);
 
 	let personalization = find_by_field(scenarios, "/scenario_id", "personalization")?;
 
@@ -3927,12 +4075,24 @@ fn assert_dreaming_readiness_baseline_counts(ledger: &Value, stages: &[Value]) -
 
 	assert_eq!(
 		consolidation.pointer("/comparison_judgment").and_then(Value::as_str),
-		Some("not_tested")
+		Some("improved")
 	);
 	assert_eq!(
 		consolidation.pointer("/baseline_counts/not_encoded").and_then(Value::as_u64),
 		Some(1)
 	);
+	assert_eq!(consolidation.pointer("/post_stage_counts/pass").and_then(Value::as_u64), Some(4));
+	assert_eq!(
+		consolidation.pointer("/post_stage_counts/not_encoded").and_then(Value::as_u64),
+		Some(0)
+	);
+	assert!(
+		consolidation
+			.pointer("/post_stage_basis")
+			.and_then(Value::as_str)
+			.is_some_and(|basis| basis.contains("apply/defer/discard audit")
+				&& basis.contains("zero source mutations"))
+	);
 
 	let scheduled = find_by_field(stages, "/stage_id", "scheduled_memory_task_readiness")?;
 
@@ -3948,6 +4108,7 @@ fn assert_dreaming_readiness_baseline_counts(ledger: &Value, stages: &[Value]) -
 	assert_eq!(retest.pointer("/baseline_counts/not_encoded").and_then(Value::as_u64), Some(11));
 	assert!(array_contains_str(ledger, "/summary/improved", "current_vs_historical_correctness")?);
 	assert!(array_contains_str(ledger, "/summary/improved", "preference_evolution")?);
+	assert!(array_contains_str(ledger, "/summary/improved", "reviewable_consolidation")?);
 	assert!(array_at(ledger, "/summary/regressed")?.is_empty());
 	assert!(array_contains_str(ledger, "/summary/unchanged", "deletion_ttl_tombstone_behavior")?);
 	assert!(array_contains_str(ledger, "/summary/unchanged", "final_competitor_retest_status")?);
@@ -3959,15 +4120,18 @@ fn assert_dreaming_readiness_baseline_counts(ledger: &Value, stages: &[Value]) -
 
 fn assert_dreaming_readiness_markdown_boundaries(markdown: &str) {
 	assert!(
-		markdown.contains("`improved`: current-vs-historical correctness and preference evolution")
+		markdown
+			.contains("`improved`: current-vs-historical correctness, preference evolution, and")
+			&& markdown.contains("reviewable consolidation")
 	);
 	assert!(markdown.contains("`regressed`: none"));
 	assert!(markdown.contains("the XY-905 run passes all six memory-evolution jobs"));
 	assert!(markdown.contains("XY-905"));
 	assert!(
 		markdown
-			.contains("Do not claim this ledger fixes preference history against mem0/OpenMemory")
+			.contains("Do not claim this ledger proves preference history against mem0/OpenMemory")
 	);
+	assert!(markdown.contains("Reviewable consolidation now has ELF live service-backed"));
 }
 
 #[test]
diff --git a/docs/guide/benchmarking/2026-06-11-competitor-strength-adoption-report.md b/docs/guide/benchmarking/2026-06-11-competitor-strength-adoption-report.md
index fee7cda8..686ed123 100644
--- a/docs/guide/benchmarking/2026-06-11-competitor-strength-adoption-report.md
+++ b/docs/guide/benchmarking/2026-06-11-competitor-strength-adoption-report.md
@@ -112,7 +112,7 @@ results, or lifecycle failures into one aggregate leaderboard.
 | Retrieval quality | `tie` | `fixture_backed`, `live_real_world`, `live_baseline_only` | ELF and qmd both pass encoded live retrieval and stress/same-corpus retrieval evidence. | XY-923 |
 | Retrieval quality and local debug UX | `loss` | `live_baseline_only`, `research_gate`, `wrong_result`, `not_encoded` | The XY-923 trace/replay report scores qmd stronger on immediate top-10 candidate artifacts and short CLI replay commands. ELF keeps useful service trace/admin replay surfaces, and expansion, fusion, rerank-on, and candidate-drop diagnostics remain untested. | XY-923 |
 | Memory evolution and temporal history | `loss` | `fixture_backed`, `live_real_world`, `live_baseline_only`, `wrong_result`, `blocked` | ELF fixture memory evolution passes, but live ELF passes only delete/TTL and reports five wrong_result jobs where current-vs-historical state is not reconciled. The mem0 local OSS preference-correction history scenario is now measured and is also an ELF loss. | XY-905 |
-| Consolidation/proposal review | `not_tested` | `fixture_backed`, `not_encoded` | ELF fixture consolidation passes, but live consolidation proposal generation and review-action scoring are not encoded. | XY-926 |
+| Consolidation/proposal review | `not_tested` for direct competitors; ELF self-check passes | `fixture_backed`, `live_real_world`, `research_gate`, `not_encoded` | ELF fixture consolidation passes and XY-934 adds live service-backed proposal materialization, lineage, confidence/usefulness, unsupported-claim flags, and apply/defer/discard audit evidence. Managed dreaming and Always-On Memory Agent patterns remain product references, not direct live competitors. | XY-934 |
 | Knowledge page compilation | `not_tested` | `fixture_backed`, `live_real_world`, `wrong_result`, `research_gate`, `blocked`, `not_encoded` | ELF fixture knowledge pages pass, but live knowledge compilation is not encoded. The XY-929 graph/RAG representative slice scores graphify as wrong_result and keeps GraphRAG, llm-wiki, and gbrain as blocked or not_tested references. | XY-926, XY-929 |
 | Operator debugging/viewer UX | `win` | `fixture_backed`, `live_real_world`, `blocked`, `not_encoded` | ELF now has a narrow live operator-debug win over qmd on trace hydration, candidate-drop visibility, and selected-but-not-narrated evidence. ELF ties qmd on replay-command availability and repair-action clarity. XY-925 adds claude-mem progressive-disclosure and retrieval-repair prompt coverage, but claude-mem viewer/operator workflows and OpenMemory UI/export remain blocked, so this is not a broad viewer-product superiority claim. | XY-926 |
 | Capture/write policy and redaction | `not_tested` | `fixture_backed`, `live_real_world`, `live_baseline_only`, `blocked`, `not_encoded` | ELF live capture/write-policy self-check jobs pass for redaction, exclusions, source ids, evidence binding, and no secret leakage. qmd remains `not_encoded`; agentmemory and claude-mem hook-capture comparisons remain `blocked` until Docker-contained hook observations and write-policy/viewer readback artifacts exist, so no broad capture-hook superiority claim is allowed. | XY-933, XY-925 |
@@ -131,7 +131,8 @@ results, or lifecycle failures into one aggregate leaderboard.
 | XY-923 | P0 | Backlog | qmd trace-level replay and wrong-result diagnostics. |
 | XY-924/XY-931 | P0 | Encoded local OSS history; UI/export setup blocker measured | mem0/OpenMemory local OSS history and SDK export-style readback are measured; OpenMemory UI/export has a blocked export-helper setup probe and still needs a dedicated compose/import path before any product-UX comparison. |
 | XY-925 | P1 | Fixture slice encoded; runtime paths still blocked | First-generation OSS prompt coverage and typed blockers are recorded for agentmemory, memsearch, and claude-mem; durable agentmemory hooks and claude-mem viewer/operator runs still need runtime adapters. |
-| XY-926 | P1 | Backlog | Live consolidation and knowledge-page suites; broad operator-debugging remains dependent on OpenMemory and claude-mem UI runners. |
+| XY-926 | P1 | Partial live suites encoded | ELF live knowledge-page scoring is encoded; broader knowledge-page external comparisons and broad operator-debugging remain dependent on contained llm-wiki/gbrain/GraphRAG/OpenMemory/claude-mem runners. Consolidation is split to XY-934. |
+| XY-934 | P1 | ELF live self-check encoded | Live consolidation proposal scoring is encoded for ELF with lineage, confidence/usefulness, unsupported-claim flags, and review-action audit; direct competitor runners remain untested or product-reference only. |
 | XY-933 | P1 | Live ELF self-check encoded | Capture/write-policy redaction, exclusion, source-id, evidence-binding, and no-leak scoring for ELF; durable agentmemory/claude-mem capture-hook comparison remains blocked. |
 | XY-927 | P1 | Fixture encoded; Letta export blocked | ELF core-vs-archival fixture coverage is encoded; a contained Letta export/readback adapter remains future work before win/tie/loss claims. |
 | XY-928 | P1 | Encoded blocked fixtures | OpenViking context-trajectory and hierarchy benchmark is encoded but blocked until evidence-bearing same-corpus and staged artifacts exist. |
diff --git a/docs/guide/benchmarking/2026-06-11-competitor-strength-evidence-matrix.md b/docs/guide/benchmarking/2026-06-11-competitor-strength-evidence-matrix.md
index c1ca8dcf..0a956467 100644
--- a/docs/guide/benchmarking/2026-06-11-competitor-strength-evidence-matrix.md
+++ b/docs/guide/benchmarking/2026-06-11-competitor-strength-evidence-matrix.md
@@ -103,7 +103,7 @@ lifecycle-fail -> `lifecycle_fail`, and not-encoded -> `not_encoded`.
 | Project decisions | Fixture and live project_decisions pass; the ELF core-archival fixture also scores project-decision recovery through core routing plus archival rationale. | qmd, Letta. | qmd live project_decisions pass; Letta project-decision recovery is `research_gate` `not_tested` or `blocked` until the contained export path exists. | Run the Letta core/archival export/readback contract before treating project-decision recovery as a comparable scenario. |
 | Source-of-truth | Fixture and live trust_source_of_truth pass. | memsearch. | memsearch canonical-store, reindex, delete, and reload smoke passes; XY-925 fixture-backed source-of-truth prompts now cover the canonical Markdown rebuild/reload boundary, but no live memsearch prompt adapter pass is claimed. | Promote memsearch source-of-truth rebuild/reload prompts into a live adapter before any suite-level win/loss claim. |
 | Temporal/current-vs-historical memory | Fixture memory_evolution passes; live memory_evolution is `wrong_result`. | Graphiti/Zep, mem0/OpenMemory. | Graphiti/Zep is `research_gate` `blocked`; mem0/OpenMemory local OSS preference history, entity scope, deletion audit, and SDK `get_all` now pass; OpenMemory UI/export is blocked by the export-helper setup probe; graph-memory scenarios are `not_encoded`. | Fix ELF/qmd live memory_evolution evidence links, add OpenMemory product app import/export readback, and run XY-888. |
-| Consolidation | Fixture consolidation passes; live consolidation is `not_encoded`. | agentmemory, managed-memory references, llm-wiki. | No manifest project has live consolidation scoring. | Run reviewable consolidation proposal generation with source refs, unsupported-claim flags, and audit transitions. |
+| Consolidation | Fixture consolidation passes; XY-934 adds ELF live service-backed proposal scoring with lineage, confidence/usefulness, unsupported-claim flags, and apply/defer/discard audit. | managed dreaming, Always-On Memory Agent patterns, agentmemory, llm-wiki. | No direct live competitor runner emits comparable consolidation artifacts; qmd remains `not_encoded`. | Keep competitor comparisons reference-only until a contained runner emits source ids, confidence, unsupported-claim flags, and review-action audit artifacts. |
 | Knowledge pages | Fixture knowledge_compilation passes; live knowledge_compilation is `not_encoded`. | llm-wiki, gbrain, GraphRAG, graphify. | llm-wiki and gbrain are `research_gate` `not_encoded` or `blocked`; GraphRAG is `blocked`; graphify has a tiny scored smoke `wrong_result`. | Encode live derived-page rebuild/lint scoring and run contained knowledge/RAG adapters only after setup proof. |
 | Operator debugging | Fixture operator_debugging_ux passes, and the narrow live operator-debug slice passes for trace hydration, candidate-drop visibility, selected-but-not-narrated evidence, replay-command availability, and repair-action clarity. | qmd, claude-mem, OpenMemory. | qmd ties replay-command availability and repair-action clarity but is `wrong_result` for trace hydration, candidate-drop stage visibility, and selected-but-not-narrated evidence. XY-925 adds claude-mem progressive-disclosure and retrieval-repair prompt coverage, while claude-mem viewer/operator and OpenMemory UI/export remain blocked. | Add bounded OpenMemory and claude-mem UI/export or viewer runners before any broader operator-UX claim. |
 | Capture/write policy | Fixture capture_integration passes; ELF live capture_integration passes 4/4 with zero redaction leaks, source ids, write-policy audit, and evidence binding. | agentmemory, claude-mem. | agentmemory and claude-mem hook capture remain `blocked` until Docker-contained hook observations and write-policy/viewer readback artifacts exist. | Run durable agentmemory and claude-mem capture-hook jobs proving redaction, exclusion, evidence binding, source ids, and no secret leakage. |
diff --git a/docs/guide/benchmarking/2026-06-11-elf-iteration-direction-from-competitor-benchmarks.md b/docs/guide/benchmarking/2026-06-11-elf-iteration-direction-from-competitor-benchmarks.md
index 6fa05a45..f5a2ad4b 100644
--- a/docs/guide/benchmarking/2026-06-11-elf-iteration-direction-from-competitor-benchmarks.md
+++ b/docs/guide/benchmarking/2026-06-11-elf-iteration-direction-from-competitor-benchmarks.md
@@ -79,11 +79,12 @@ Interpretation:
 - Both pass `trust_source_of_truth`, `work_resume`, `project_decisions`,
   `retrieval`, and `personalization`.
 - Both fail most `memory_evolution` live conflict evidence with `wrong_result`.
-- ELF now passes live `capture_integration`; qmd keeps that suite `not_encoded`.
-  Both leave consolidation, knowledge compilation, and production-ops operator
-  boundaries as `not_encoded` or `blocked`. Operator debugging has a separate narrow
-  live slice: ELF passes it, while qmd remains `wrong_result` for trace hydration and
-  candidate-drop stage visibility.
+- ELF now passes live `capture_integration`. A separate XY-934 narrow run adds live
+  consolidation proposal review evidence for ELF; qmd keeps consolidation
+  `not_encoded` in the live sweep. Knowledge compilation and production-ops operator
+  boundaries remain typed `not_encoded` or `blocked`. Operator debugging has a
+  separate narrow live slice: ELF passes it, while qmd remains `wrong_result` for
+  trace hydration and candidate-drop stage visibility.
 
 ### Production Evidence
 
@@ -134,7 +135,7 @@ one misleading score.
 | Project decisions | ELF and qmd live project-decision suites pass; ELF fixture-backed `core_archival_memory` also scores project-decision recovery, while Letta remains blocked without export evidence. | Run the Letta core/archival export/readback contract before treating project-decision recovery as comparable. |
 | Source of truth | ELF has the strongest measured source-of-truth evidence. | Borrow memsearch's local canonical-store ergonomics without making files or vectors authoritative. |
 | Temporal memory | ELF fixture passes, but live memory evolution is wrong_result. | Prioritize current-vs-historical evidence links and Graphiti/Zep-style validity windows. |
-| Consolidation | ELF fixture passes, but live proposal generation is not encoded. | Build reviewable derived proposals with source refs, confidence, unsupported-claim flags, and apply/defer/discard audit. |
+| Consolidation | ELF fixture passes and XY-934 adds live service-backed proposal materialization, lineage, confidence/usefulness, unsupported-claim flags, and apply/defer/discard audit; direct competitor runners remain untested. | Keep derived proposal review as the safety boundary and add competitor/reference runners only when they emit comparable artifacts. |
 | Knowledge pages | ELF fixture pages pass; live knowledge generation is not encoded. | Borrow llm-wiki lint/query-save loops, gbrain timelines, and graphify reports behind rebuild/lint benchmarks. |
 | Operator debugging | Fixture UX passes and the narrow live trace/viewer slice is scored: ELF passes, qmd ties replay/repair clarity but is wrong_result for trace hydration and candidate-drop visibility. | Expand coverage to OpenMemory and claude-mem UI/export or viewer runners before any broader operator-UX claim. |
 | Capture/write policy | ELF live capture/write-policy self-check passes with zero redaction leaks; qmd is `not_encoded`; agentmemory is `blocked`; claude-mem is `not_encoded`. | Borrow agentmemory/claude-mem capture breadth only after durable local hook/viewer evidence exists, while preserving redaction and evidence binding. |
@@ -213,9 +214,13 @@ These improve day-to-day usefulness while preserving ELF's evidence-bound core.
 
 2. Reviewable consolidation
    - Borrow from: managed memory dreaming and Always-On Memory Agent scheduling.
+   - Current state: ELF now has live service-backed proposal scoring for the
+     consolidation fixture slice; direct competitor/reference runners are still
+     untested.
    - ELF shape: derived proposals only; source notes are not silently rewritten.
-   - Benchmark gate: consolidation proposals include lineage, confidence,
-     unsupported-claim flags, and apply/defer/discard audit.
+   - Benchmark gate: preserve lineage, confidence, unsupported-claim flags,
+     apply/defer/discard audit, and zero source mutations; do not add scheduling until
+     it can remain derived and reviewable.
 
 3. Knowledge pages
    - Borrow from: llm-wiki, gbrain, graphify, and GraphRAG.
diff --git a/docs/guide/benchmarking/2026-06-11-measurement-coverage-audit.md b/docs/guide/benchmarking/2026-06-11-measurement-coverage-audit.md
index 470a89a7..841e945f 100644
--- a/docs/guide/benchmarking/2026-06-11-measurement-coverage-audit.md
+++ b/docs/guide/benchmarking/2026-06-11-measurement-coverage-audit.md
@@ -102,6 +102,11 @@ live adapter or competitor runtime can complete those jobs.
 
 `cargo make real-world-memory-live-adapters` produced:
 
+XY-934 update: the June 11 consolidation row below is superseded for ELF by
+`docs/guide/benchmarking/2026-06-16-live-consolidation-proposal-scoring-report.md`.
+ELF now has live service-backed consolidation proposal scoring for the 4 checked-in
+consolidation jobs; qmd remains typed `not_encoded` for this suite.
+
 | Adapter | Jobs | Pass | Wrong result | Blocked | Not encoded | Mean score | Mean latency | Evidence recall | Evidence coverage |
 | --- | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: |
 | ELF live service adapter | `40` | `22` | `5` | `2` | `11` | `0.599` | `6.980 ms` | `50/80` | `58/88` |
@@ -167,7 +172,7 @@ records `unique_project_names: 17` for the full project list including ELF.
 
 | Project | Best current evidence | Current measured state | Strongest unproven scenario | Next measurement before claim |
 | --- | --- | --- | --- | --- |
-| ELF | `fixture_backed` plus `live_real_world` | Fixture aggregate passes except 5 blocked operator or measurement-gate boundaries; live full sweep is `wrong_result`; live capture/write-policy and narrow operator-debug slices pass. | Full live memory evolution, live consolidation, live knowledge pages, live production ops, competitor capture hooks, OpenViking staged trajectory artifacts, and broader operator UI runners. | Memory-evolution diagnostic report, then consolidation/knowledge reports plus agentmemory/claude-mem capture, OpenViking staged trajectory artifacts, and OpenMemory/claude-mem UI runners. |
+| ELF | `fixture_backed` plus `live_real_world` | Fixture aggregate passes except 5 blocked operator or measurement-gate boundaries; live full sweep is `wrong_result`; live capture/write-policy, live consolidation proposal scoring, and narrow operator-debug slices pass. | Full live memory evolution, live knowledge pages, live production ops, competitor capture hooks, OpenViking staged trajectory artifacts, and broader operator UI runners. | Memory-evolution diagnostic report, then knowledge reports plus agentmemory/claude-mem capture, OpenViking staged trajectory artifacts, and OpenMemory/claude-mem UI runners. |
 | qmd | `live_real_world` plus `live_baseline_only` | Fresh full sweep is five passes behind ELF because qmd misses the delete/TTL tombstone job and keeps capture/write-policy jobs typed `not_encoded`; same-corpus baseline passes; narrow operator-debug live slice ties replay commands but is `wrong_result` for trace hydration and candidate-drop visibility. | Deep retrieval-debug ergonomics and trace replay beyond the narrow operator-debug slice. | qmd/ELF deep retrieval-debug profile with expansion, fusion, rerank, and dropped-candidate traces. |
 | agentmemory | `live_baseline_only` | `lifecycle_fail`; capture comparison is `blocked` because the Docker baseline uses a process-local StateKV Map and in-memory index, with no durable local session/capture path for source ids, exclusions, write-policy audit, or evidence-bound output. | Durable coding-agent continuity and capture hooks. | Durable lifecycle and work-resume/capture adapter report. |
 | mem0/OpenMemory | `live_baseline_only` | Basic local smoke and local OSS history/readback pass; OpenMemory UI/export is blocked, hosted Platform export is a non-goal, and optional graph plus broader prompt coverage remain `not_encoded`. | Entity history, lifecycle UI, OpenMemory inspection. | Entity-history, deletion-audit, and UI/export readback report. |
@@ -194,7 +199,7 @@ records `unique_project_names: 17` for the full project list including ELF.
 | Project decisions | ELF and qmd live pass; ELF fixture coverage also passes core routing plus archival rationale recovery. | ELF is credible on encoded project-decision recovery. | Letta core/archival decision memory export and scoring. |
 | Source of truth | ELF and qmd live pass; ELF has stronger production restore/rebuild evidence. | ELF has strongest measured source-of-truth discipline. | memsearch source-of-truth reindex/reload evidence. |
 | Memory evolution | ELF live fails 5/6 jobs; qmd live fails 6/6 jobs after missing the delete/TTL tombstone evidence; fixture aggregate passes. | No broad live superiority claim. | Historical conflict evidence links and Graphiti/Zep temporal comparison. |
-| Consolidation | Fixture aggregate passes; live adapters are not encoded. | Fixture-only claim. | Live proposal generation with lineage, confidence, and review-action audit. |
+| Consolidation | Fixture aggregate passes; XY-934 adds ELF live service-backed proposal scoring, while qmd remains `not_encoded`. | ELF self-check claim only; no direct competitor win. | Contained competitor/reference runners only when they emit source ids, confidence, unsupported-claim flags, and review-action audit. |
 | Knowledge pages | Fixture aggregate passes; live adapters are not encoded. | Fixture-only claim. | Live page rebuild/lint plus llm-wiki, gbrain, GraphRAG, and graphify comparisons. |
 | Operator debugging | Fixture aggregate passes; narrow ELF/qmd live operator-debug slice is scored with ELF `pass` and qmd `wrong_result`. | Narrow ELF/qmd live claim only: ELF wins trace hydration, candidate-drop visibility, and selected-but-not-narrated evidence; replay-command and repair-action clarity are tied. | OpenMemory and claude-mem UI/export or viewer runners before any broader operator-UX claim. |
 | Capture/write policy | Fixture aggregate passes; ELF live service adapter passes 4/4 capture jobs with zero redaction leaks; qmd is `not_encoded`; agentmemory is `blocked`; claude-mem hook/viewer capture is `blocked`. | ELF has live self-check evidence for redaction, exclusions, source ids, evidence binding, and no secret leakage. Against agentmemory/claude-mem capture breadth, the comparison remains blocked until durable hook/viewer evidence exists. | Durable agentmemory and claude-mem capture-hook runners with evidence-bound output. |
diff --git a/docs/guide/benchmarking/2026-06-16-dreaming-readiness-stage-ledger.md b/docs/guide/benchmarking/2026-06-16-dreaming-readiness-stage-ledger.md
index 0239e21c..df37634e 100644
--- a/docs/guide/benchmarking/2026-06-16-dreaming-readiness-stage-ledger.md
+++ b/docs/guide/benchmarking/2026-06-16-dreaming-readiness-stage-ledger.md
@@ -18,15 +18,16 @@ This ledger does not claim a broad product win. It records the gate later produc
 lanes must pass before they can claim a Dreaming or competitor-inspired stage is done,
 and now includes the XY-905 post-stage result for live temporal reconciliation.
 
-Current baseline:
+Current stage status:
 
-- `improved`: current-vs-historical correctness and preference evolution.
+- `improved`: current-vs-historical correctness, preference evolution, and
+  reviewable consolidation.
 - `regressed`: none.
 - `unchanged`: deletion/TTL/tombstone behavior and the final competitor retest
   baseline.
 - `blocked`: scheduled-memory-task readiness.
-- `not_tested`: reviewable consolidation beyond fixtures, memory-summary/top-of-mind
-  live behavior, and proactive brief readiness.
+- `not_tested`: memory-summary/top-of-mind live behavior and proactive brief
+  readiness.
 
 The known live `memory_evolution` loss is now repaired for the encoded ELF live
 adapter slice: the XY-905 run passes all six memory-evolution jobs and reports
@@ -34,6 +35,11 @@ current, historical, rationale, tombstone, invalidation, selected, dropped, and
 non-narrated evidence fields. This is not a private-corpus, hosted memory, or broad
 competitor-superiority claim.
 
+Reviewable consolidation is also improved for the narrow ELF self-check: XY-934 adds
+service-backed proposal materialization, source lineage, confidence/usefulness,
+unsupported-claim flags, apply/defer/discard audit transitions, and zero source
+mutations. Direct competitor runners remain untested or product-reference only.
+
 ## Ledger Rules
 
 - Every downstream Dreaming or competitor-improvement stage must write a post-stage
@@ -57,7 +63,7 @@ competitor-superiority claim.
 | Current-vs-historical correctness | `cargo make real-world-memory-evolution`; `cargo make real-world-memory-live-adapters` | Same commands; publish post-stage JSON and Markdown evidence | `pass=1`, `wrong_result=5`, `blocked=0`, `not_tested=0`, `not_encoded=0` | `pass=6`, `wrong_result=0`, `blocked=0`, `not_tested=0`, `not_encoded=0` | `improved` | Move from benchmark materialization into service-native temporal reconciliation APIs and compare against mem0/OpenMemory history and Graphiti/Zep temporal graph evidence without broad superiority claims. |
 | Preference evolution and correction history | `cargo make real-world-memory-evolution`; `cargo make real-world-memory-live-adapters`; `cargo make openmemory-ui-export-readback` | Same commands; include mem0/OpenMemory boundary evidence | `pass=0`, `wrong_result=1`, `blocked=0`, `not_tested=0`, `not_encoded=0` | `pass=1`, `wrong_result=0`, `blocked=0`, `not_tested=0`, `not_encoded=0` | `improved` | Measure preference correction against mem0/OpenMemory history and UI/export surfaces before making any broader history-quality claim. |
 | Deletion, TTL, and tombstone behavior | `cargo make real-world-memory`; `cargo make real-world-memory-live-adapters` | Same commands | `pass=1`, `wrong_result=0`, `blocked=0`, `not_tested=0`, `not_encoded=0` | `pass=1`, `wrong_result=0`, `blocked=0`, `not_tested=0`, `not_encoded=0` | `unchanged` | Extend tombstone and TTL readback beyond the single encoded job into update/delete/recreate history cases. |
-| Reviewable consolidation | `cargo make real-world-memory-consolidation` | `cargo make real-world-memory-consolidation`; `cargo make real-world-memory-live-adapters` | `pass=4`, `wrong_result=0`, `blocked=0`, `not_tested=1`, `not_encoded=1` | not run by XY-905 | `not_tested` | Keep Dreaming output derived and reviewable with lineage, confidence, unsupported-claim flags, apply/defer/discard audit, and no source mutation. |
+| Reviewable consolidation | `cargo make real-world-memory-consolidation` | `cargo make real-world-memory-consolidation`; `cargo make real-world-memory-live-consolidation`; `cargo make real-world-memory-live-adapters` | `pass=4`, `wrong_result=0`, `blocked=0`, `not_tested=1`, `not_encoded=1` | `pass=4`, `wrong_result=0`, `blocked=0`, `not_tested=0`, `not_encoded=0` | `improved` | Keep Dreaming output derived and reviewable, and add direct competitor/reference runners only when they emit comparable source ids, confidence, unsupported-claim flags, and review audit artifacts. |
 | Memory summary and top-of-mind behavior | `cargo make real-world-memory-knowledge`; `cargo make real-world-memory-core-archival` | Same commands plus `cargo make real-world-memory-live-adapters` | `pass=8`, `wrong_result=0`, `blocked=0`, `not_tested=1`, `not_encoded=1` | not run by XY-905 | `not_tested` | Build summaries as cited, rebuildable derived pages or core blocks; do not turn hidden summaries into authoritative memory. |
 | Proactive brief readiness | `cargo make real-world-first-generation-oss`; `cargo make real-world-job-operator-ux` | Same commands plus `cargo make real-world-memory-live-adapters` | `pass=0`, `wrong_result=0`, `blocked=0`, `not_tested=1`, `not_encoded=1` | not run by XY-905 | `not_tested` | Add direct proactive-brief fixtures before any pass claim; briefs must be source-linked and repairable. |
 | Scheduled memory task readiness | `cargo make real-world-memory-consolidation` | `cargo make real-world-memory-consolidation`; `cargo make real-world-memory-live-adapters` | `pass=0`, `wrong_result=0`, `blocked=1`, `not_tested=0`, `not_encoded=0` | not run by XY-905 | `blocked` | Scheduled runs are future work; start with queued derived proposal runs and keep operator review mandatory. |
@@ -70,7 +76,7 @@ competitor-superiority claim.
 | Current-vs-historical correctness | `docs/guide/benchmarking/2026-06-16-live-temporal-reconciliation-report.md`; `docs/research/2026-06-16-live-temporal-reconciliation-report.json`; `docs/guide/benchmarking/2026-06-11-temporal-history-competitor-gap-report.md`; `docs/research/2026-06-11-temporal-history-competitor-gap-report.json`; `docs/guide/benchmarking/2026-06-11-competitor-strength-adoption-report.md` |
 | Preference evolution and correction history | `docs/guide/benchmarking/2026-06-16-live-temporal-reconciliation-report.md`; `docs/research/2026-06-16-live-temporal-reconciliation-report.json`; `docs/guide/benchmarking/2026-06-11-temporal-history-competitor-gap-report.md`; `docs/guide/benchmarking/2026-06-11-mem0-openmemory-history-ui-export-report.md`; `docs/research/2026-06-11-temporal-history-competitor-gap-report.json` |
 | Deletion, TTL, and tombstone behavior | `docs/guide/benchmarking/2026-06-16-live-temporal-reconciliation-report.md`; `docs/research/2026-06-16-live-temporal-reconciliation-report.json`; `docs/guide/benchmarking/2026-06-11-temporal-history-competitor-gap-report.md`; `docs/guide/benchmarking/2026-06-11-measurement-coverage-audit.md` |
-| Reviewable consolidation | `docs/spec/system_consolidation_proposals_v1.md`; `apps/elf-eval/fixtures/real_world_memory/consolidation/`; `docs/guide/benchmarking/2026-06-11-competitor-strength-adoption-report.md` |
+| Reviewable consolidation | `docs/spec/system_consolidation_proposals_v1.md`; `apps/elf-eval/fixtures/real_world_memory/consolidation/`; `docs/guide/benchmarking/2026-06-16-live-consolidation-proposal-scoring-report.md`; `docs/research/2026-06-16-live-consolidation-proposal-scoring-report.json` |
 | Memory summary and top-of-mind behavior | `apps/elf-eval/fixtures/real_world_memory/knowledge/`; `apps/elf-eval/fixtures/real_world_memory/core_archival_memory/`; `docs/guide/benchmarking/2026-06-11-competitor-strength-adoption-report.md` |
 | Proactive brief readiness | `docs/research/2026-06-08-agent-memory-selection.json`; `docs/guide/benchmarking/2026-06-11-first-generation-oss-continuity-source-store-report.md` |
 | Scheduled memory task readiness | `docs/spec/system_consolidation_proposals_v1.md`; `docs/research/2026-06-08-agent-memory-selection.json` |
@@ -103,14 +109,16 @@ Allowed:
   files.
 - The current ledger preserves typed non-pass states and records the XY-905 live
   memory-evolution improvement.
-- Fixture-backed consolidation, knowledge, and core/archival jobs can be used as
-  regression guards for report shape.
+- Fixture-backed knowledge and core/archival jobs can be used as regression guards for
+  report shape.
+- Reviewable consolidation now has ELF live service-backed proposal scoring evidence,
+  with direct competitor runners still untested.
 
 Not allowed:
 
-- Do not claim this ledger fixes preference history against mem0/OpenMemory,
-  consolidation, proactive briefs, scheduled tasks, private-corpus gates, hosted
-  memory, or competitor adapters.
+- Do not claim this ledger proves preference history against mem0/OpenMemory,
+  proactive briefs, scheduled tasks, private-corpus gates, hosted memory, broad
+  consolidation superiority, or competitor adapters.
 - Do not claim ELF has full-suite live real-world pass evidence.
 - Do not claim private-corpus or provider-backed production quality without the
   operator-owned inputs required by XY-930.
diff --git a/docs/guide/benchmarking/2026-06-16-live-consolidation-proposal-scoring-report.md b/docs/guide/benchmarking/2026-06-16-live-consolidation-proposal-scoring-report.md
new file mode 100644
index 00000000..4e7f8302
--- /dev/null
+++ b/docs/guide/benchmarking/2026-06-16-live-consolidation-proposal-scoring-report.md
@@ -0,0 +1,86 @@
+# Live Consolidation Proposal Scoring Report - June 16, 2026
+
+Goal: Record the XY-934 live consolidation proposal scoring evidence and product
+reference boundaries.
+Read this when: You need to know whether ELF has live evidence for reviewable
+consolidation proposal generation, source lineage, confidence, unsupported-claim
+flags, and apply/defer/discard review audit transitions.
+Inputs: `cargo make real-world-memory-consolidation`,
+`cargo make real-world-memory-live-consolidation`,
+`apps/elf-eval/fixtures/real_world_memory/consolidation/`,
+`apps/elf-eval/src/bin/real_world_live_adapter.rs`, and
+`docs/spec/system_consolidation_proposals_v1.md`.
+Outputs: Scenario-level consolidation results, live artifacts, and typed comparison
+boundaries for managed dreaming and Always-On Memory Agent style references.
+
+## Verdict
+
+ELF now has service-backed live consolidation proposal scoring. The narrow live
+command materializes all 4 `consolidation` jobs through `ElfService` consolidation
+run creation, worker proposal materialization, and review-action audit transitions.
+
+This is not scheduled production consolidation and not live provider generation. The
+run uses the deterministic fixture/manual proposal payload boundary required by
+`elf.consolidation/v1`: source notes are immutable, proposals are derived outputs, and
+review actions are explicit artifacts.
+
+## Fresh Runs
+
+| Command | Result | Artifact |
+| --- | --- | --- |
+| `cargo make real-world-memory-consolidation` | pass | `tmp/real-world-memory/consolidation/report.json` |
+| `cargo make real-world-memory-live-consolidation` | pass | `tmp/real-world-memory/live-consolidation/summary.json` |
+
+## ELF Live Consolidation Results
+
+| Job | Live status | Source refs | Review action | Final review state | Unsupported claims | Source mutations |
+| --- | --- | ---: | --- | --- | ---: | ---: |
+| `consolidation-project-summary-apply-001` | `pass` | `2` | `apply` | `applied` | `0` | `0` |
+| `consolidation-weekly-decision-summary-apply-001` | `pass` | `2` | `apply` | `applied` | `0` | `0` |
+| `consolidation-preference-candidate-defer-001` | `pass` | `2` | `defer` | `archived` | `0` | `0` |
+| `consolidation-contradiction-report-discard-001` | `pass` | `3` | `discard` | `rejected` | `1` | `0` |
+
+The generated benchmark report keeps the same consolidation metrics as the fixture
+report:
+
+- `proposal_count = 4`
+- `lineage_completeness = 1.0`
+- `review_action_correctness = 1.0`
+- `proposal_unsupported_claim_count = 1`
+- `source_mutation_count = 0`
+- `executable_gap_count = 0`
+
+The materialization artifact records service-backed run ids, proposal ids, source
+lineage counts, unsupported-claim flag counts, review-event counts, review actions,
+and final review states. It does not claim source memory rewrites.
+
+## Comparison Boundary
+
+| Compared target | Position | Reason |
+| --- | --- | --- |
+| qmd live real-world adapter | `untested` | qmd keeps consolidation jobs typed `not_encoded`; no qmd consolidation proposal generator or review-action audit runner exists in this benchmark. |
+| Managed dreaming memory systems | `product_reference` | Managed dreaming motivates the proposal-review shape, but no contained runner emits comparable source ids, confidence, unsupported-claim flags, and review audit artifacts. |
+| Always-On Memory Agent patterns | `product_reference` | Always-on scheduling remains a reference only. XY-934 does not implement scheduled consolidation and does not allow silent source-of-truth rewrites. |
+
+## Claims Allowed
+
+- ELF live consolidation self-checks pass for proposal materialization, source
+  lineage, confidence/usefulness thresholds, unsupported-claim flags, and
+  apply/defer/discard audit transitions.
+- Fixture consolidation passes and live service-backed consolidation evidence are
+  separate evidence classes.
+- qmd and other tracked projects remain untested or reference-only for live
+  consolidation proposal scoring until a contained runner emits comparable artifacts.
+- Derived-output safety claims are tied to source lineage, immutable source snapshots,
+  zero source mutations, and review-action artifacts.
+
+## Claims Not Allowed
+
+- Do not claim scheduled production consolidation exists.
+- Do not claim live provider-generated consolidation quality; the accepted
+  `elf.consolidation/v1` service boundary is deterministic fixture/manual proposal
+  materialization.
+- Do not claim ELF broadly beats managed dreaming, Always-On Memory Agent,
+  agentmemory, qmd, or llm-wiki on consolidation without comparable contained live
+  runners.
+- Do not mix knowledge-page rebuild/lint scoring into the consolidation claim.
diff --git a/docs/guide/benchmarking/index.md b/docs/guide/benchmarking/index.md
index 21f9b7b8..c6d926a5 100644
--- a/docs/guide/benchmarking/index.md
+++ b/docs/guide/benchmarking/index.md
@@ -105,6 +105,10 @@ cleanup, use `docs/guide/single_user_production.md`.
   report that scores ELF redaction, exclusions, source ids, evidence binding, and no
   secret leakage while preserving typed blocked/untested boundaries for agentmemory
   and claude-mem capture breadth.
+- `2026-06-16-live-consolidation-proposal-scoring-report.md`: XY-934 live
+  consolidation proposal scoring report that separates fixture-backed consolidation
+  passes from service-backed live proposal materialization, lineage, confidence,
+  unsupported-claim flags, and apply/defer/discard audit evidence.
 - `2026-06-11-mem0-openmemory-history-ui-export-report.md`: XY-924 plus XY-931
   mem0/OpenMemory local OSS history, preference-correction, deletion-audit,
   personalization, and export-readback comparison with normalized
diff --git a/docs/guide/benchmarking/real_world_agent_memory_benchmark.md b/docs/guide/benchmarking/real_world_agent_memory_benchmark.md
index 81693524..ce1bcc1d 100644
--- a/docs/guide/benchmarking/real_world_agent_memory_benchmark.md
+++ b/docs/guide/benchmarking/real_world_agent_memory_benchmark.md
@@ -402,6 +402,27 @@ These fixtures use the same reviewable proposal shape as the runtime manual/fixt
 consolidation service. They remain offline fixture responses and do not claim scheduled
 provider-backed proposal generation.
 
+Current live consolidation increment:
+
+```sh
+cargo make real-world-memory-live-consolidation
+```
+
+This runs only `apps/elf-eval/fixtures/real_world_memory/consolidation/` through the
+ELF live service adapter and writes:
+
+```text
+tmp/real-world-memory/live-consolidation/elf-materialization.json
+tmp/real-world-memory/live-consolidation/elf-report.json
+tmp/real-world-memory/live-consolidation/elf-report.md
+tmp/real-world-memory/live-consolidation/summary.json
+```
+
+The live increment proves service-backed proposal materialization and review audit for
+the current checked-in consolidation jobs. It does not implement scheduled production
+consolidation, live provider-generated proposal quality, source-of-truth rewrites, or
+knowledge-page rebuild/lint scoring.
+
 Current checked-in knowledge-compilation increment:
 
 ```sh
diff --git a/docs/research/2026-06-11-competitor-strength-adoption-report.json b/docs/research/2026-06-11-competitor-strength-adoption-report.json
index c918eab9..bc5761b4 100644
--- a/docs/research/2026-06-11-competitor-strength-adoption-report.json
+++ b/docs/research/2026-06-11-competitor-strength-adoption-report.json
@@ -229,16 +229,19 @@
       "outcome": "not_tested",
       "evidence_classes": [
         "fixture_backed",
+        "live_real_world",
+        "research_gate",
         "not_encoded"
       ],
-      "measured_claim": "ELF fixture consolidation passes, but live consolidation proposal generation and review-action scoring are not encoded.",
+      "measured_claim": "ELF fixture consolidation passes, and XY-934 adds live service-backed proposal materialization, source lineage, confidence/usefulness, unsupported-claim flags, and apply/defer/discard audit evidence. Managed dreaming and Always-On Memory Agent patterns remain product references, not direct live competitors.",
       "command_artifacts": [
-        "docs/guide/benchmarking/2026-06-11-measurement-coverage-audit.md"
+        "docs/guide/benchmarking/2026-06-16-live-consolidation-proposal-scoring-report.md",
+        "docs/research/2026-06-16-live-consolidation-proposal-scoring-report.json"
       ],
       "follow_up_issues": [
-        "XY-926"
+        "XY-934"
       ],
-      "caveat": "Fixture evidence cannot be promoted into live proposal-quality proof."
+      "caveat": "The live evidence is an ELF self-check for deterministic fixture/manual proposal materialization; no direct managed dreaming, Always-On Memory Agent, qmd, agentmemory, or llm-wiki live competitor runner is claimed."
     },
     {
       "scenario_id": "knowledge_page_compilation",
diff --git a/docs/research/2026-06-11-xy-897-competitor-strength-matrix.json b/docs/research/2026-06-11-xy-897-competitor-strength-matrix.json
index 59e5a19f..3de690bd 100644
--- a/docs/research/2026-06-11-xy-897-competitor-strength-matrix.json
+++ b/docs/research/2026-06-11-xy-897-competitor-strength-matrix.json
@@ -478,11 +478,11 @@
     {
       "scenario_id": "consolidation",
       "scenario": "consolidation",
-      "current_elf_evidence": "ELF fixture-backed consolidation passes, but live_real_world consolidation is not_encoded.",
-      "strongest_competitor_or_reference": "agentmemory, managed dreaming references, llm-wiki",
-      "current_competitor_evidence": "Manifest projects do not yet have live consolidation scoring; llm-wiki knowledge workflow is research_gate not_encoded.",
-      "current_state": "Fixture-only ELF evidence is useful, but no live proposal-generation parity claim is allowed.",
-      "next_measurement": "Run a reviewable consolidation-worker benchmark that emits proposals, source refs, unsupported-claim flags, and apply/discard/defer audit events."
+      "current_elf_evidence": "ELF fixture-backed consolidation passes, and XY-934 adds live_real_world service-backed proposal scoring with source lineage, confidence/usefulness, unsupported-claim flags, apply/defer/discard audit, and zero source mutations.",
+      "strongest_competitor_or_reference": "managed dreaming, Always-On Memory Agent patterns, agentmemory, llm-wiki",
+      "current_competitor_evidence": "No direct live competitor runner emits comparable consolidation artifacts; qmd remains not_encoded and managed dreaming plus Always-On Memory Agent patterns are product references only.",
+      "current_state": "ELF has live consolidation self-check evidence, but no broad consolidation superiority or direct competitor parity claim is allowed without contained external runners.",
+      "next_measurement": "Add contained competitor/reference runners only if they can emit source ids, confidence, unsupported-claim flags, and review-action audit artifacts."
     },
     {
       "scenario_id": "knowledge_pages",
diff --git a/docs/research/2026-06-16-dreaming-readiness-stage-ledger.json b/docs/research/2026-06-16-dreaming-readiness-stage-ledger.json
index 596791e9..76104dc5 100644
--- a/docs/research/2026-06-16-dreaming-readiness-stage-ledger.json
+++ b/docs/research/2026-06-16-dreaming-readiness-stage-ledger.json
@@ -4,7 +4,7 @@
   "authority": "XY-951",
   "created_at": "2026-06-16T00:00:00Z",
   "purpose": "Define the benchmark evidence gate that every Dreaming-inspired ELF optimization stage must update before claiming completion.",
-  "source_evidence_cutoff": "Checked-in benchmark and research evidence through the XY-905 live temporal reconciliation run on 2026-06-16; no private-corpus or provider-backed production pass is claimed by this ledger.",
+  "source_evidence_cutoff": "Checked-in benchmark and research evidence through the XY-905 live temporal reconciliation run and XY-934 live consolidation proposal scoring run on 2026-06-16; no private-corpus or provider-backed production pass is claimed by this ledger.",
   "typed_status_terms": [
     "pass",
     "wrong_result",
@@ -36,12 +36,14 @@
     "Typed non-pass states must remain typed; blocked, not_tested, not_encoded, incomplete, lifecycle_fail, unsupported, and wrong_result must not be collapsed into a generic fail or hidden under pass.",
     "Fixture-backed evidence may prove benchmark shape but must not be promoted into live_real_world product quality.",
     "Private-corpus and provider-backed production gates remain typed blocked unless the operator supplies explicit inputs; those blockers are tracked under XY-930.",
-    "The XY-905 post-stage live memory_evolution result is a narrow temporal reconciliation improvement only; it must not be converted into private-corpus, hosted memory, or broad competitor superiority claims."
+    "The XY-905 post-stage live memory_evolution result is a narrow temporal reconciliation improvement only; it must not be converted into private-corpus, hosted memory, or broad competitor superiority claims.",
+    "The XY-934 live consolidation result is a narrow ELF self-check only; it must not be converted into broad managed dreaming, Always-On Memory Agent, qmd, agentmemory, or llm-wiki superiority claims without comparable contained runners."
   ],
   "summary": {
     "improved": [
       "current_vs_historical_correctness",
-      "preference_evolution"
+      "preference_evolution",
+      "reviewable_consolidation"
     ],
     "regressed": [],
     "unchanged": [
@@ -52,7 +54,6 @@
       "scheduled_memory_task_readiness"
     ],
     "not_tested": [
-      "reviewable_consolidation",
       "memory_summary_top_of_mind_behavior",
       "proactive_brief_readiness"
     ]
@@ -234,8 +235,8 @@
     {
       "stage_id": "reviewable_consolidation",
       "stage_name": "Reviewable consolidation",
-      "dependent_issue": "XY-926",
-      "evidence_class": "fixture_backed",
+      "dependent_issue": "XY-934",
+      "evidence_class": "live_real_world",
       "baseline_commands": [
         {
           "command": "cargo make real-world-memory-consolidation",
@@ -248,6 +249,10 @@
           "command": "cargo make real-world-memory-consolidation",
           "required_artifact": "tmp/real-world-memory/consolidation/report.json"
         },
+        {
+          "command": "cargo make real-world-memory-live-consolidation",
+          "required_artifact": "tmp/real-world-memory/live-consolidation/summary.json"
+        },
         {
           "command": "cargo make real-world-memory-live-adapters",
           "required_artifact": "tmp/real-world-memory/live-adapters/"
@@ -255,7 +260,8 @@
       ],
       "evidence_files": [
         "docs/spec/system_consolidation_proposals_v1.md",
-        "docs/guide/benchmarking/2026-06-11-competitor-strength-adoption-report.md",
+        "docs/guide/benchmarking/2026-06-16-live-consolidation-proposal-scoring-report.md",
+        "docs/research/2026-06-16-live-consolidation-proposal-scoring-report.json",
         "apps/elf-eval/fixtures/real_world_memory/consolidation/"
       ],
       "baseline_counts": {
@@ -265,11 +271,19 @@
         "not_tested": 1,
         "not_encoded": 1
       },
-      "baseline_basis": "Consolidation fixtures pass, but live consolidation proposal generation and review-action scoring are not encoded.",
-      "comparison_judgment": "not_tested",
+      "baseline_basis": "Before XY-934, consolidation fixtures passed but live consolidation proposal generation and review-action scoring were not encoded.",
+      "post_stage_counts": {
+        "pass": 4,
+        "wrong_result": 0,
+        "blocked": 0,
+        "not_tested": 0,
+        "not_encoded": 0
+      },
+      "post_stage_basis": "XY-934 adds ELF live service-backed proposal materialization, source lineage, confidence/usefulness, unsupported-claim flags, apply/defer/discard audit, and zero source mutations for 4 consolidation jobs.",
+      "comparison_judgment": "improved",
       "regression_rule": "Any source mutation, missing lineage, or collapse of review actions into an automatic rewrite is a regression.",
-      "improvement_rule": "An improvement requires live or service-backed consolidation scoring without provider hidden state and without mutating authoritative sources.",
-      "next_optimization_direction": "Keep Dreaming output derived and reviewable: proposal lineage, confidence, unsupported-claim flags, apply/defer/discard audit, and immutable source snapshots."
+      "improvement_rule": "The stage is improved when live or service-backed consolidation scoring exists without provider hidden state and without mutating authoritative sources.",
+      "next_optimization_direction": "Keep Dreaming output derived and reviewable, and add direct competitor/reference runners only when they emit comparable source ids, confidence, unsupported-claim flags, and review audit artifacts."
     },
     {
       "stage_id": "memory_summary_top_of_mind_behavior",
diff --git a/docs/research/2026-06-16-live-consolidation-proposal-scoring-report.json b/docs/research/2026-06-16-live-consolidation-proposal-scoring-report.json
new file mode 100644
index 00000000..4f33fed9
--- /dev/null
+++ b/docs/research/2026-06-16-live-consolidation-proposal-scoring-report.json
@@ -0,0 +1,137 @@
+{
+  "schema": "elf.live_consolidation_proposal_scoring_report/v1",
+  "report_id": "xy-934-live-consolidation-proposal-scoring-2026-06-16",
+  "authority": "XY-934",
+  "created_at": "2026-06-16T00:00:00Z",
+  "commands": [
+    {
+      "command": "cargo make real-world-memory-consolidation",
+      "status": "pass",
+      "artifact": "tmp/real-world-memory/consolidation/report.json"
+    },
+    {
+      "command": "cargo make real-world-memory-live-consolidation",
+      "status": "pass",
+      "artifact": "tmp/real-world-memory/live-consolidation/summary.json"
+    }
+  ],
+  "fixture_aggregate": {
+    "suite_id": "consolidation",
+    "evidence_class": "fixture_backed",
+    "encoded_job_count": 4,
+    "suite_status": "pass",
+    "proposal_count": 4,
+    "source_mutation_count": 0,
+    "proposal_unsupported_claim_count": 1,
+    "lineage_completeness": 1.0,
+    "review_action_correctness": 1.0,
+    "executable_gap_count": 0
+  },
+  "live_consolidation_results": {
+    "elf_live_real_world": {
+      "evidence_class": "live_real_world",
+      "suite_status": "pass",
+      "encoded_job_count": 4,
+      "proposal_count": 4,
+      "source_mutation_count": 0,
+      "proposal_unsupported_claim_count": 1,
+      "lineage_completeness": 1.0,
+      "review_action_correctness": 1.0,
+      "review_event_count": 6,
+      "artifact": "tmp/real-world-memory/live-consolidation/elf-report.json",
+      "materialization_artifact": "tmp/real-world-memory/live-consolidation/elf-materialization.json"
+    },
+    "qmd_live_real_world": {
+      "evidence_class": "live_real_world",
+      "suite_status": "not_encoded",
+      "encoded_job_count": 4,
+      "proposal_count": 0,
+      "artifact": "tmp/real-world-memory/live-adapters/qmd-report.json"
+    }
+  },
+  "jobs": [
+    {
+      "job_id": "consolidation-project-summary-apply-001",
+      "status": "pass",
+      "proposal_kind": "project_summary",
+      "source_lineage_count": 2,
+      "usefulness_score": 0.93,
+      "min_usefulness_score": 0.8,
+      "review_action": "apply",
+      "final_review_state": "applied",
+      "review_event_count": 2,
+      "unsupported_claim_flag_count": 0,
+      "source_mutation_count": 0
+    },
+    {
+      "job_id": "consolidation-weekly-decision-summary-apply-001",
+      "status": "pass",
+      "proposal_kind": "weekly_decision_summary",
+      "source_lineage_count": 2,
+      "usefulness_score": 0.91,
+      "min_usefulness_score": 0.8,
+      "review_action": "apply",
+      "final_review_state": "applied",
+      "review_event_count": 2,
+      "unsupported_claim_flag_count": 0,
+      "source_mutation_count": 0
+    },
+    {
+      "job_id": "consolidation-preference-candidate-defer-001",
+      "status": "pass",
+      "proposal_kind": "preference_candidate",
+      "source_lineage_count": 2,
+      "usefulness_score": 0.86,
+      "min_usefulness_score": 0.75,
+      "review_action": "defer",
+      "final_review_state": "archived",
+      "review_event_count": 1,
+      "unsupported_claim_flag_count": 0,
+      "source_mutation_count": 0
+    },
+    {
+      "job_id": "consolidation-contradiction-report-discard-001",
+      "status": "pass",
+      "proposal_kind": "contradiction_report",
+      "source_lineage_count": 3,
+      "usefulness_score": 0.9,
+      "min_usefulness_score": 0.8,
+      "review_action": "discard",
+      "final_review_state": "rejected",
+      "review_event_count": 1,
+      "unsupported_claim_flag_count": 1,
+      "source_mutation_count": 0
+    }
+  ],
+  "reference_positions": [
+    {
+      "project": "qmd",
+      "position": "untested",
+      "reason": "qmd keeps consolidation jobs typed not_encoded in the full live sweep; no proposal generation or review-action audit runner exists for qmd."
+    },
+    {
+      "project": "managed_dreaming_memory_systems",
+      "position": "product_reference",
+      "reason": "Managed dreaming motivates the derived proposal-review shape, but no contained runner emits comparable source ids, confidence, unsupported-claim flags, and review audit artifacts."
+    },
+    {
+      "project": "always_on_memory_agent_patterns",
+      "position": "product_reference",
+      "reason": "Always-on scheduling remains a reference only; XY-934 does not implement scheduled consolidation and does not allow silent source-of-truth rewrites."
+    }
+  ],
+  "claim_boundary": {
+    "allowed": [
+      "ELF live consolidation self-checks pass for proposal materialization, source lineage, confidence/usefulness thresholds, unsupported-claim flags, and apply/defer/discard audit transitions.",
+      "Fixture consolidation passes and live service-backed consolidation evidence are separate evidence classes.",
+      "qmd and other tracked projects remain untested or reference-only for live consolidation proposal scoring until a contained runner emits comparable artifacts.",
+      "Derived-output safety claims are tied to source lineage, immutable source snapshots, zero source mutations, and review-action artifacts."
+    ],
+    "not_allowed": [
+      "Do not claim scheduled production consolidation exists.",
+      "Do not claim live provider-generated consolidation quality; the accepted elf.consolidation/v1 service boundary is deterministic fixture/manual proposal materialization.",
+      "Do not claim ELF broadly beats managed dreaming, Always-On Memory Agent, agentmemory, qmd, or llm-wiki on consolidation without comparable contained live runners.",
+      "Do not mix knowledge-page rebuild/lint scoring into the consolidation claim."
+    ]
+  }
+}
diff --git a/scripts/real-world-consolidation-live-adapter.sh b/scripts/real-world-consolidation-live-adapter.sh
new file mode 100755
index 00000000..5d506134
--- /dev/null
+++ b/scripts/real-world-consolidation-live-adapter.sh
@@ -0,0 +1,69 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+ROOT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
+REPORT_DIR="${ELF_CONSOLIDATION_LIVE_REPORT_DIR:-${ROOT_DIR}/tmp/real-world-memory/live-consolidation}"
+FIXTURE_DIR="${ELF_CONSOLIDATION_LIVE_FIXTURES:-${ROOT_DIR}/apps/elf-eval/fixtures/real_world_memory/consolidation}"
+
+if [[ ! -f "/.dockerenv" && "${ELF_CONSOLIDATION_LIVE_ALLOW_HOST:-0}" != "1" ]]; then
+  echo "Refusing to run live consolidation adapter outside Docker. Use cargo make real-world-memory-live-consolidation." >&2
+  exit 1
+fi
+
+for cmd in bash cargo jq; do
+  if ! command -v "${cmd}" >/dev/null 2>&1; then
+    echo "Missing ${cmd} in live consolidation runner." >&2
+    exit 1
+  fi
+done
+
+mkdir -p "${REPORT_DIR}"
+rm -rf "${REPORT_DIR:?}/elf-fixtures" \
+  "${REPORT_DIR:?}/elf-materialization.json" \
+  "${REPORT_DIR:?}/elf-report.json" \
+  "${REPORT_DIR:?}/elf-report.md" \
+  "${REPORT_DIR:?}/summary.json"
+
+cd "${ROOT_DIR}"
+
+cargo run -p elf-eval --bin real_world_live_adapter -- elf \
+  --fixtures "${FIXTURE_DIR}" \
+  --out-fixtures "${REPORT_DIR}/elf-fixtures" \
+  --evidence-out "${REPORT_DIR}/elf-materialization.json" \
+  --config config/local/elf.docker.toml
+
+cargo run -p elf-eval --bin real_world_job_benchmark -- run \
+  --fixtures "${REPORT_DIR}/elf-fixtures" \
+  --out "${REPORT_DIR}/elf-report.json" \
+  --run-id real-world-memory-live-consolidation \
+  --adapter-id elf_live_real_world \
+  --adapter-name "ELF live consolidation service adapter" \
+  --adapter-behavior live_real_world_adapter \
+  --adapter-storage-status pass \
+  --adapter-runtime-status pass \
+  --adapter-notes "Materialized by real_world_live_adapter through ElfService consolidation_run_create, worker proposal materialization, and apply/defer/discard review audit transitions; source notes remain immutable derived-output evidence."
+
+cargo run -p elf-eval --bin real_world_job_benchmark -- publish \
+  --report "${REPORT_DIR}/elf-report.json" \
+  --out "${REPORT_DIR}/elf-report.md"
+
+jq -n \
+  --slurpfile materialization "${REPORT_DIR}/elf-materialization.json" \
+  --slurpfile report "${REPORT_DIR}/elf-report.json" \
+  '{
+    schema: "elf.real_world_consolidation_live_adapter_sweep/v1",
+    generated_at: (now | todateiso8601),
+    fixture_dir: (env.ELF_CONSOLIDATION_LIVE_FIXTURES // "apps/elf-eval/fixtures/real_world_memory/consolidation"),
+    artifact_dir: (env.ELF_CONSOLIDATION_LIVE_REPORT_DIR // "tmp/real-world-memory/live-consolidation"),
+    adapter: {
+      adapter_id: "elf_live_real_world",
+      evidence_class: "live_real_world",
+      materialization: $materialization[0],
+      report: {
+        json: "tmp/real-world-memory/live-consolidation/elf-report.json",
+        markdown: "tmp/real-world-memory/live-consolidation/elf-report.md",
+        summary: $report[0].summary,
+        suites: $report[0].suites
+      }
+    }
+  }' >"${REPORT_DIR}/summary.json"