CaspianG · CaspianG · Jul 11, 2026 · Jul 11, 2026 · Jul 11, 2026 · Jul 11, 2026
diff --git a/README.md b/README.md
diff --git a/benchmarks/AGENT_IMPACT.md b/benchmarks/AGENT_IMPACT.md
@@ -1,6 +1,6 @@
 # WaveMind Agent Impact Leaderboard
 
-Generated: `2026-07-11T03:37:56Z`.
+Generated: `2026-07-11T05:56:44Z`.
 
 Agent-impact rows come from checked-in benchmark artifacts. They show behavioral lift on the configured tasks; they do not claim general agent success outside the listed scenarios.
 

diff --git a/benchmarks/BENCHMARK_LEADERBOARD.md b/benchmarks/BENCHMARK_LEADERBOARD.md
@@ -1,7 +1,7 @@
 # WaveMind Benchmark Leaderboard
 
 Generated from `benchmarks/benchmark_matrix_results.json`.
-Last refresh: `2026-07-11T03:37:56Z` from `be2cebfd776b`.
+Last refresh: `2026-07-11T05:56:44Z` from `c4f786e131c8`.
 
 This is a compact reader-facing view of checked-in benchmark results. It is not a universal vector-database leaderboard: each row uses the primary quality metric for that benchmark, and latency is shown separately so quality wins are not confused with speed wins.
 
@@ -34,7 +34,7 @@ This is a compact reader-facing view of checked-in benchmark results. It is not
 
 | area | current source | claim status | next action |
 |---|---|---|---|
-| Artifact freshness | local matrix refresh at `2026-07-11T03:37:56Z` | source `be2cebfd776b`; audit gate enforced by `validate_benchmark_artifacts.py` | Keep weekly refresh green before public claims. |
+| Artifact freshness | local matrix refresh at `2026-07-11T05:56:44Z` | source `c4f786e131c8`; audit gate enforced by `validate_benchmark_artifacts.py` | Keep weekly refresh green before public claims. |
 | Serverless telemetry | loopback API pool; `loopback-api-capacity-estimate`; 4 measured replicas | observed SLO `True`; loopback evidence, not a managed-serverless claim | Run `.github/workflows/serverless-observed-telemetry.yml` against deployed API nodes. |
 | External HTTP cluster load | kubernetes-kind-non-loopback-ci; `kubernetes-pod-dns-physical-node-drill`; 4 nodes | SLO `True`; non-loopback Kubernetes pod-DNS evidence | Run `.github/workflows/external-http-cluster-load.yml` with a remote node manifest. |
 | External HTTP active-active loopback | local-loopback; `loopback-api-regions`; 3 regions | SLO `True`; external URL contract over local API regions | Run `.github/workflows/external-http-active-active.yml` with remote regions for production evidence. |
@@ -43,7 +43,7 @@ This is a compact reader-facing view of checked-in benchmark results. It is not
 | 10M streaming load | local `WaveMind faiss-ivfpq-persisted streaming` profile | target recall `0.99`, p99 `60.1 ms`, SLO `scale_required` | Repeat at 50M and add service-backed Qdrant/pgvector 10M artifacts. |
 | 50M streaming preflight | `WaveMind faiss-ivfpq-persisted streaming` plan-only contract | `action_required`; index `1.12 GB`; app storage `119.2 GB`; blockers `missing_env:WAVEMIND_FAISS_IVFPQ_PATH, insufficient_local_disk_for_index_and_transient_batches` | Run `.github/workflows/production-streaming-load.yml` with `faiss-ivfpq-persisted` and publish `benchmarks/production_streaming_load_ivfpq_50m_results.json`. |
 | Qdrant streaming | real Qdrant service smoke plus measured 10M profile | smoke recall `1`, smoke p99 `17.9 ms`; 10M recall `0.975`, 10M p99 `43.3 ms`, SLO `scale_required` | Keep the measured 10M profile green and run the sharded Qdrant and pgvector 10M profiles next. |
-| Qdrant sharded streaming | real two-service fanout smoke plus horizontal Qdrant preflight | smoke recall `1`, smoke p99 `16.0 ms`; 10M preflight `action_required`; 100M preflight `action_required`; planned shards `4`; blockers `missing_env:WAVEMIND_QDRANT_URLS, insufficient_local_disk_for_index_and_transient_batches` | Run `.github/workflows/production-streaming-load.yml` with `qdrant-sharded-service` and publish `benchmarks/production_streaming_load_qdrant_sharded_10m_results.json` or `benchmarks/production_streaming_load_qdrant_sharded_100m_results.json`. |
+| Qdrant sharded streaming | real fanout smoke plus measured four-service 10M profile | smoke recall `1`, smoke p99 `16.0 ms`; 10M recall `0.993`, 10M p99 `71.3 ms`, shards `4`; 100M preflight `action_required`; planned shards `4`; blockers `none (measured artifact passes)` | Keep the measured 10M sharded profile green and run the strict 100M sharded profile next. |
 | Qdrant 1M streaming | real Qdrant service run before and after warmup/chunking tuning | cold p99 `3014.0 ms`; tuned recall `1`, tuned p99 `26.4 ms`, SLO `pass` | Use the tuned warmup/chunking profile for the 10M Qdrant service run. |
 | pgvector streaming | real PostgreSQL/pgvector service smoke plus 10M preflight | smoke recall `1`, smoke p99 `7.624 ms`; 10M preflight `action_required` | Run `.github/workflows/production-streaming-load.yml` with `pgvector-service` against sized Postgres storage. |
 | Production readiness gate | checked-in benchmark artifacts | `pass`; 39/39 pass | Keep the gate at readiness_score 1.0 while repeating larger service-backed runs and moving external competitor evidence into the separate adapter profile. |

diff --git a/benchmarks/BENCHMARK_REPORT.md b/benchmarks/BENCHMARK_REPORT.md
diff --git a/benchmarks/COST_EFFICIENCY.md b/benchmarks/COST_EFFICIENCY.md
@@ -1,16 +1,16 @@
 # WaveMind Cost Efficiency Leaderboard
 
-Generated: `2026-07-11T03:37:56Z`.
+Generated: `2026-07-11T05:56:44Z`.
 
 Measured rows come from checked-in load artifacts. Planned rows are capacity and cost contracts only; they do not unlock production latency or recall claims until the matching benchmark result exists.
 
 ## Summary
 
-- Measured rows: `21`.
+- Measured rows: `22`.
 - Measured SLO pass rows: `9`.
-- Measured valid cost rows: `12`.
+- Measured valid cost rows: `13`.
 - Planned cost rows: `5`.
-- Measured frontier: `sub_100k-wavemind-numpy-streaming-production_streaming_load_smoke_results, 100k-qdrant-service-production_load_qdrant_100k_tuned_results, 1m-qdrant-service-streaming-production_streaming_load_qdrant_1m_tuned_results, 1m-wavemind-faiss-ivfpq-persisted-streaming-production_streaming_load_ivfpq_1m_results, 100k-wavemind-faiss-ivfpq-persisted-streaming-production_streaming_load_ivfpq_100k_results, 10m-wavemind-faiss-ivfpq-persisted-streaming-production_streaming_load_ivfpq_10m_results, 10m-qdrant-service-streaming-production_streaming_load_qdrant_10m_results`.
+- Measured frontier: `sub_100k-wavemind-numpy-streaming-production_streaming_load_smoke_results, 100k-qdrant-service-production_load_qdrant_100k_tuned_results, 1m-qdrant-service-streaming-production_streaming_load_qdrant_1m_tuned_results, 1m-wavemind-faiss-ivfpq-persisted-streaming-production_streaming_load_ivfpq_1m_results, 100k-wavemind-faiss-ivfpq-persisted-streaming-production_streaming_load_ivfpq_100k_results, 10m-qdrant-sharded-service-streaming-production_streaming_load_qdrant_sharded_10m_results, 10m-wavemind-faiss-ivfpq-persisted-streaming-production_streaming_load_ivfpq_10m_results, 10m-qdrant-service-streaming-production_streaming_load_qdrant_10m_results`.
 - Planned frontier: `qdrant-sharded-100m, faiss-ivfpq-50m`.
 
 ## Measured Cost Frontier
@@ -27,8 +27,8 @@ Measured rows come from checked-in load artifacts. Planned rows are capacity and
 | 8 | 1m-wavemind-faiss-ivfpq-persisted-streaming-production_streaming_load_ivfpq_1m_results | 1m | WaveMind faiss-ivfpq-persisted streaming | 1,000,000 | 0.99 | 4.992 | pass | $0.694 | $182.74 | `benchmarks/production_streaming_load_ivfpq_1m_results.json` |
 | 9 | 100k-wavemind-faiss-ivfpq-persisted-streaming-production_streaming_load_ivfpq_100k_results | 100k | WaveMind faiss-ivfpq-persisted streaming | 100,000 | 0.96 | 1.104 | pass | $0.694 | $182.52 | `benchmarks/production_streaming_load_ivfpq_100k_results.json` |
 | 10 | 1m-wavemind-faiss-persisted-production_load_faiss_1m_results | 1m | WaveMind faiss-persisted | 1,000,000 | 1 | 57.71 | scale_required | $4.167 | $1,095.24 | `benchmarks/production_load_faiss_1m_results.json` |
-| 11 | 10m-wavemind-faiss-ivfpq-persisted-streaming-production_streaming_load_ivfpq_10m_results | 10m | WaveMind faiss-ivfpq-persisted streaming | 10,000,000 | 0.99 | 60.13 | scale_required | $4.861 | $1,279.88 | `benchmarks/production_streaming_load_ivfpq_10m_results.json` |
-| 12 | 10m-qdrant-service-streaming-production_streaming_load_qdrant_10m_results | 10m | Qdrant service streaming | 10,000,000 | 0.975 | 43.27 | scale_required | $3.472 | $914.88 | `benchmarks/production_streaming_load_qdrant_10m_results.json` |
+| 11 | 10m-qdrant-sharded-service-streaming-production_streaming_load_qdrant_sharded_10m_results | 10m | Qdrant sharded service streaming | 10,000,000 | 0.993 | 71.28 | scale_required | $4.722 | $3,104.88 | `benchmarks/production_streaming_load_qdrant_sharded_10m_results.json` |
+| 12 | 10m-wavemind-faiss-ivfpq-persisted-streaming-production_streaming_load_ivfpq_10m_results | 10m | WaveMind faiss-ivfpq-persisted streaming | 10,000,000 | 0.99 | 60.13 | scale_required | $4.861 | $1,279.88 | `benchmarks/production_streaming_load_ivfpq_10m_results.json` |
 
 ## Planned Cost Frontier
 

diff --git a/benchmarks/PRODUCTION_EVIDENCE.md b/benchmarks/PRODUCTION_EVIDENCE.md
@@ -7,8 +7,8 @@ multi-region, managed-serverless, 50M, or 100M production scale cannot.
 | metric | value |
 |---|---:|
 | overall status | `action_required` |
-| passed requirements | `3` |
-| action required | `5` |
+| passed requirements | `4` |
+| action required | `4` |
 | failed requirements | `0` |
 | total requirements | `8` |
 
@@ -18,7 +18,7 @@ multi-region, managed-serverless, 50M, or 100M production scale cannot.
 | External HTTP active-active regions | `action_required` | no checked-in external HTTP active-active region result; issues: missing artifact | `benchmarks/external_http_active_active_results.json` | `gh workflow run external-http-active-active.yml -f regions="us-east=https://wm-us.example.com,eu-west=https://wm-eu.example.com,ap-south=https://wm-ap.example.com" -f namespace_count=16 -f p99_slo_ms=1500 -f fail_on_slo=true -f commit_results=true` | Remote multi-region active-active memory convergence. |
 | Managed/serverless remote telemetry | `action_required` | missing remote serverless telemetry; issues: missing artifact | `deploy/serverless/observed-telemetry.remote.json` | `gh workflow run serverless-observed-telemetry.yml -f nodes="https://wm-a.example.com,https://wm-b.example.com" -f seed_mode=first -f commit_results=true` | Hosted/serverless p99, cold-start, error-rate, and scale-out SLO. |
 | 10M Qdrant service load | `pass` | Qdrant service streaming: vectors 10000000, recall 0.975, p99 43.266599997878075 ms, cost valid_slo, source be2cebfd776b, run local-qdrant10m-quantized-ef1024-v1182-20260711 | `benchmarks/production_streaming_load_qdrant_10m_results.json` | `python benchmarks/production_streaming_load_benchmark.py --sizes 10000000 --dim 128 --queries 2000 --top-k 10 --seed 42 --noise 0.08 --batch-size 5000 --engines qdrant-service --target-recall 0.95 --target-p99-ms 100.0 --target-qps 100.0 --replicas 3 --autoscaling-max-replicas 24 --capacity-headroom 0.7 --output benchmarks/production_streaming_load_qdrant_10m_results.json --checkpoint-path state/production-runs/qdrant-service-10000000.checkpoint.json` | 10M Qdrant service-backed candidate index SLO. |
-| 10M sharded Qdrant service load | `action_required` | no checked-in 10,000,000-vector result for Qdrant sharded service streaming; issues: missing artifact | `benchmarks/production_streaming_load_qdrant_sharded_10m_results.json` | `python benchmarks/production_streaming_load_benchmark.py --sizes 10000000 --dim 128 --queries 2000 --top-k 10 --seed 42 --noise 0.08 --batch-size 5000 --engines qdrant-sharded-service --target-recall 0.95 --target-p99-ms 100.0 --target-qps 250.0 --replicas 4 --autoscaling-max-replicas 48 --capacity-headroom 0.7 --output benchmarks/production_streaming_load_qdrant_sharded_10m_results.json --checkpoint-path state/production-runs/qdrant-sharded-service-10000000.checkpoint.json` | Horizontally sharded Qdrant service recall/latency SLO. |
+| 10M sharded Qdrant service load | `pass` | Qdrant sharded service streaming: vectors 10000000, recall 0.9925, p99 71.27849999233149 ms, cost valid_slo, source 11d0be8c62e8, run local-qdrant-sharded10m-4x-v1182-20260711 | `benchmarks/production_streaming_load_qdrant_sharded_10m_results.json` | `python benchmarks/production_streaming_load_benchmark.py --sizes 10000000 --dim 128 --queries 2000 --top-k 10 --seed 42 --noise 0.08 --batch-size 5000 --engines qdrant-sharded-service --target-recall 0.95 --target-p99-ms 100.0 --target-qps 250.0 --replicas 4 --autoscaling-max-replicas 48 --capacity-headroom 0.7 --output benchmarks/production_streaming_load_qdrant_sharded_10m_results.json --checkpoint-path state/production-runs/qdrant-sharded-service-10000000.checkpoint.json` | Horizontally sharded Qdrant service recall/latency SLO. |
 | 10M pgvector service load | `action_required` | no checked-in 10,000,000-vector result for WaveMind pgvector streaming; issues: missing artifact | `benchmarks/production_streaming_load_pgvector_10m_results.json` | `python benchmarks/production_streaming_load_benchmark.py --sizes 10000000 --dim 128 --queries 2000 --top-k 10 --seed 42 --noise 0.08 --batch-size 5000 --engines pgvector-service --target-recall 0.95 --target-p99-ms 100.0 --target-qps 100.0 --replicas 3 --autoscaling-max-replicas 24 --capacity-headroom 0.7 --output benchmarks/production_streaming_load_pgvector_10m_results.json --checkpoint-path state/production-runs/pgvector-service-10000000.checkpoint.json` | 10M PostgreSQL/pgvector service candidate-index SLO. |
 | 50M FAISS IVF-PQ streaming load | `pass` | WaveMind faiss-ivfpq-persisted streaming: vectors 50000000, recall 0.9705, p99 73.10794899996154 ms, cost valid_slo, source a9ea9d7f8027, run github-actions-29067655616 | `benchmarks/production_streaming_load_ivfpq_50m_results.json` | `python benchmarks/production_streaming_load_benchmark.py --sizes 50000000 --dim 128 --queries 2000 --top-k 10 --seed 42 --noise 0.08 --batch-size 1000000 --engines faiss-ivfpq-persisted --target-recall 0.95 --target-p99-ms 100.0 --target-qps 100.0 --replicas 3 --autoscaling-max-replicas 24 --capacity-headroom 0.7 --output benchmarks/production_streaming_load_ivfpq_50m_results.json --checkpoint-path state/production-runs/faiss-ivfpq-persisted-50000000.checkpoint.json` | 50M compressed local/persistent FAISS profile. |
 | 100M remote load result | `action_required` | no checked-in 100,000,000-vector result for Qdrant sharded service streaming; issues: missing artifact | `benchmarks/production_streaming_load_qdrant_sharded_100m_results.json` | `python benchmarks/production_streaming_load_benchmark.py --sizes 100000000 --dim 128 --queries 5000 --top-k 10 --seed 42 --noise 0.08 --batch-size 10000 --engines qdrant-sharded-service --target-recall 0.95 --target-p99-ms 100.0 --target-qps 500.0 --replicas 8 --autoscaling-max-replicas 128 --capacity-headroom 0.7 --output benchmarks/production_streaming_load_qdrant_sharded_100m_results.json --checkpoint-path state/production-runs/qdrant-sharded-service-100000000.checkpoint.json` | 100M+ memories with measured sharded-Qdrant recall, p99, and cost SLO. |
diff --git a/benchmarks/PRODUCTION_EVIDENCE_BUNDLE.md b/benchmarks/PRODUCTION_EVIDENCE_BUNDLE.md
@@ -7,7 +7,7 @@ claim boundaries, and the exact next actions required to unlock blocked claims.
 | metric | value |
 |---|---:|
 | claim status | `claims_limited` |
-| strict evidence | `3/8` |
+| strict evidence | `4/8` |
 | preflight ready | `0/8` |
 | production readiness | `pass` |
 | readiness score | `1.0` |
@@ -16,7 +16,7 @@ claim boundaries, and the exact next actions required to unlock blocked claims.
 | production scale run contract | `available` |
 | production scale profiles | `5` |
 | production scale target memories | `180000000` |
-| next actions | `5` |
+| next actions | `4` |
 
 ## Claim Boundaries
 
@@ -44,6 +44,5 @@ claim boundaries, and the exact next actions required to unlock blocked claims.
 |---|---|---|---|---|---|
 | External HTTP active-active regions | `action_required` | `action_required` | `benchmarks/external_http_active_active_results.json` | `WAVEMIND_ACTIVE_ACTIVE_REGIONS, WAVEMIND_ACTIVE_ACTIVE_REGIONS_MANIFEST_JSON; issues: missing artifact` | `gh workflow run external-http-active-active.yml -f regions="$WAVEMIND_ACTIVE_ACTIVE_REGIONS" -f commit_results=true` |
 | Managed/serverless remote telemetry | `action_required` | `action_required` | `deploy/serverless/observed-telemetry.remote.json` | `WAVEMIND_SERVERLESS_NODES; issues: missing artifact` | `gh workflow run serverless-observed-telemetry.yml -f nodes="$WAVEMIND_SERVERLESS_NODES" -f seed_mode=first -f commit_results=true` |
-| 10M sharded Qdrant service load | `action_required` | `action_required` | `benchmarks/production_streaming_load_qdrant_sharded_10m_results.json` | `WAVEMIND_QDRANT_URLS; issues: missing artifact` | `python benchmarks/production_streaming_load_benchmark.py --sizes 10000000 --dim 128 --queries 2000 --top-k 10 --seed 42 --noise 0.08 --batch-size 5000 --engines qdrant-sharded-service --target-recall 0.95 --target-p99-ms 100.0 --target-qps 250.0 --replicas 4 --autoscaling-max-replicas 48 --capacity-headroom 0.7 --output benchmarks/production_streaming_load_qdrant_sharded_10m_results.json --checkpoint-path state/production-runs/qdrant-sharded-service-10000000.checkpoint.json` |
 | 10M pgvector service load | `action_required` | `action_required` | `benchmarks/production_streaming_load_pgvector_10m_results.json` | `WAVEMIND_PGVECTOR_DSN; issues: missing artifact` | `python benchmarks/production_streaming_load_benchmark.py --sizes 10000000 --dim 128 --queries 2000 --top-k 10 --seed 42 --noise 0.08 --batch-size 5000 --engines pgvector-service --target-recall 0.95 --target-p99-ms 100.0 --target-qps 100.0 --replicas 3 --autoscaling-max-replicas 24 --capacity-headroom 0.7 --output benchmarks/production_streaming_load_pgvector_10m_results.json --checkpoint-path state/production-runs/pgvector-service-10000000.checkpoint.json` |
 | 100M remote load result | `action_required` | `action_required` | `benchmarks/production_streaming_load_qdrant_sharded_100m_results.json` | `WAVEMIND_QDRANT_URLS; issues: missing artifact` | `python benchmarks/production_streaming_load_benchmark.py --sizes 100000000 --dim 128 --queries 5000 --top-k 10 --seed 42 --noise 0.08 --batch-size 10000 --engines qdrant-sharded-service --target-recall 0.95 --target-p99-ms 100.0 --target-qps 500.0 --replicas 8 --autoscaling-max-replicas 128 --capacity-headroom 0.7 --output benchmarks/production_streaming_load_qdrant_sharded_100m_results.json --checkpoint-path state/production-runs/qdrant-sharded-service-100000000.checkpoint.json` |
diff --git a/benchmarks/PRODUCTION_EVIDENCE_DISPATCH.md b/benchmarks/PRODUCTION_EVIDENCE_DISPATCH.md
@@ -9,8 +9,8 @@ strict production-evidence validation.
 |---|---:|
 | overall status | `action_required` |
 | ready to dispatch | `0` |
-| blocked by preflight | `5` |
-| complete | `3` |
+| blocked by preflight | `4` |
+| complete | `4` |
 | total jobs | `8` |
 | runner label | `self-hosted-large` |
 | commit results default | `False` |
@@ -23,7 +23,7 @@ strict production-evidence validation.
 | External HTTP active-active regions | `blocked_by_preflight` | `remote-service` | `external-http-active-active.yml` | `benchmarks/external_http_active_active_results.json` | `WAVEMIND_ACTIVE_ACTIVE_REGIONS, WAVEMIND_ACTIVE_ACTIVE_REGIONS_MANIFEST_JSON` |
 | Managed/serverless remote telemetry | `blocked_by_preflight` | `remote-service` | `serverless-observed-telemetry.yml` | `deploy/serverless/observed-telemetry.remote.json` | `WAVEMIND_SERVERLESS_NODES` |
 | 10M Qdrant service load | `complete` | `service-scale-10m` | `production-streaming-load.yml` | `benchmarks/production_streaming_load_qdrant_10m_results.json` | `WAVEMIND_QDRANT_URL` |
-| 10M sharded Qdrant service load | `blocked_by_preflight` | `service-scale-10m` | `production-streaming-load.yml` | `benchmarks/production_streaming_load_qdrant_sharded_10m_results.json` | `WAVEMIND_QDRANT_URLS` |
+| 10M sharded Qdrant service load | `complete` | `service-scale-10m` | `production-streaming-load.yml` | `benchmarks/production_streaming_load_qdrant_sharded_10m_results.json` | `WAVEMIND_QDRANT_URLS` |
 | 10M pgvector service load | `blocked_by_preflight` | `service-scale-10m` | `production-streaming-load.yml` | `benchmarks/production_streaming_load_pgvector_10m_results.json` | `WAVEMIND_PGVECTOR_DSN` |
 | 50M FAISS IVF-PQ streaming load | `complete` | `large-local-index` | `production-streaming-load.yml` | `benchmarks/production_streaming_load_ivfpq_50m_results.json` | `WAVEMIND_FAISS_IVFPQ_PATH` |
 | 100M remote load result | `blocked_by_preflight` | `hundred-million-service` | `production-streaming-load.yml` | `benchmarks/production_streaming_load_qdrant_sharded_100m_results.json` | `WAVEMIND_QDRANT_URLS` |