diff --git a/arc/benchmark.sh b/arc/benchmark.sh
index ddd7f28318..a33527ee30 100755
--- a/arc/benchmark.sh
+++ b/arc/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 exec ../lib/benchmark-common.sh
diff --git a/bqn/benchmark.sh b/bqn/benchmark.sh
index bad104cedb..fb3b4d1318 100755
--- a/bqn/benchmark.sh
+++ b/bqn/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/byconity/benchmark.sh b/byconity/benchmark.sh
index 618bc1fc0b..8764daabd5 100755
--- a/byconity/benchmark.sh
+++ b/byconity/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 # byconity boots a chain of containers (fdb -> tso -> server -> workers
 # / daemon-manager). Each later step waits up to 600s for its
diff --git a/cedardb-parquet/benchmark.sh b/cedardb-parquet/benchmark.sh
index 8b58049bec..8791ddc90e 100755
--- a/cedardb-parquet/benchmark.sh
+++ b/cedardb-parquet/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export PGHOST="/tmp"
 export PGUSER=postgres
diff --git a/cedardb/benchmark.sh b/cedardb/benchmark.sh
index 8b58049bec..8791ddc90e 100755
--- a/cedardb/benchmark.sh
+++ b/cedardb/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export PGHOST="/tmp"
 export PGUSER=postgres
diff --git a/chdb-dataframe/benchmark.sh b/chdb-dataframe/benchmark.sh
index 4148581913..0a7321ea9c 100755
--- a/chdb-dataframe/benchmark.sh
+++ b/chdb-dataframe/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_DURABLE=no
 # Skip the pre-snapshot ./stop+./start cycle: the loaded
diff --git a/chdb-parquet-partitioned/benchmark.sh b/chdb-parquet-partitioned/benchmark.sh
index fea177f34d..b7ae16ae72 100755
--- a/chdb-parquet-partitioned/benchmark.sh
+++ b/chdb-parquet-partitioned/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-partitioned"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/chdb/benchmark.sh b/chdb/benchmark.sh
index 1b692a2155..75c3806085 100755
--- a/chdb/benchmark.sh
+++ b/chdb/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-csv"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/chyt/benchmark.sh b/chyt/benchmark.sh
index 21a189925f..0fdf5fac69 100755
--- a/chyt/benchmark.sh
+++ b/chyt/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 # CHYT executes against a remote YT cluster ($YT_PROXY); no local download.
 export YT_USE_HOSTS=0
 export CHYT_ALIAS="${CHYT_ALIAS:-*ch_public}"
diff --git a/citus/benchmark.sh b/citus/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/citus/benchmark.sh
+++ b/citus/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/clickhouse-datalake-partitioned/benchmark.sh b/clickhouse-datalake-partitioned/benchmark.sh
index a5587169d1..d82911fa20 100755
--- a/clickhouse-datalake-partitioned/benchmark.sh
+++ b/clickhouse-datalake-partitioned/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 # Data is read directly from S3, no local download.
 export BENCH_DOWNLOAD_SCRIPT=""
 export BENCH_RESTARTABLE=no
diff --git a/clickhouse-datalake/benchmark.sh b/clickhouse-datalake/benchmark.sh
index a5587169d1..d82911fa20 100755
--- a/clickhouse-datalake/benchmark.sh
+++ b/clickhouse-datalake/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 # Data is read directly from S3, no local download.
 export BENCH_DOWNLOAD_SCRIPT=""
 export BENCH_RESTARTABLE=no
diff --git a/clickhouse-parquet-partitioned/benchmark.sh b/clickhouse-parquet-partitioned/benchmark.sh
index fea177f34d..b7ae16ae72 100755
--- a/clickhouse-parquet-partitioned/benchmark.sh
+++ b/clickhouse-parquet-partitioned/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-partitioned"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/clickhouse-parquet/benchmark.sh b/clickhouse-parquet/benchmark.sh
index bad104cedb..fb3b4d1318 100755
--- a/clickhouse-parquet/benchmark.sh
+++ b/clickhouse-parquet/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/clickhouse-tencent/benchmark.sh b/clickhouse-tencent/benchmark.sh
index 12a00b626c..d5cf12db78 100755
--- a/clickhouse-tencent/benchmark.sh
+++ b/clickhouse-tencent/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-partitioned"
 exec ../lib/benchmark-common.sh
diff --git a/clickhouse-web/benchmark.sh b/clickhouse-web/benchmark.sh
index 37571e2848..19cbb25805 100755
--- a/clickhouse-web/benchmark.sh
+++ b/clickhouse-web/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 # Data is read from a remote ClickHouse-hosted web disk; no local download.
 export BENCH_DOWNLOAD_SCRIPT=""
 exec ../lib/benchmark-common.sh
diff --git a/clickhouse/benchmark.sh b/clickhouse/benchmark.sh
index 12a00b626c..d5cf12db78 100755
--- a/clickhouse/benchmark.sh
+++ b/clickhouse/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-partitioned"
 exec ../lib/benchmark-common.sh
diff --git a/cloudberry/benchmark.sh b/cloudberry/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/cloudberry/benchmark.sh
+++ b/cloudberry/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/cockroachdb/benchmark.sh b/cockroachdb/benchmark.sh
index 0be570a340..ede6b1c438 100755
--- a/cockroachdb/benchmark.sh
+++ b/cockroachdb/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-csv"
 # cockroach replays its WAL on each restart; after the 60 GB+ IMPORT
 # that takes long enough that the lib's default 300 s check window
diff --git a/cratedb/benchmark.sh b/cratedb/benchmark.sh
index 51a5b6e329..08453e5fec 100755
--- a/cratedb/benchmark.sh
+++ b/cratedb/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 # Set CRATEDB_MODE=tuned to use create-tuned.sql + queries-tuned.sql.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 
diff --git a/daft-parquet-partitioned/benchmark.sh b/daft-parquet-partitioned/benchmark.sh
index 024c58fe16..435a4a8d76 100755
--- a/daft-parquet-partitioned/benchmark.sh
+++ b/daft-parquet-partitioned/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-partitioned"
 export BENCH_DURABLE=no
 # Skip the pre-snapshot ./stop+./start cycle: the loaded
diff --git a/daft-parquet/benchmark.sh b/daft-parquet/benchmark.sh
index 4148581913..0a7321ea9c 100755
--- a/daft-parquet/benchmark.sh
+++ b/daft-parquet/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_DURABLE=no
 # Skip the pre-snapshot ./stop+./start cycle: the loaded
diff --git a/databend/benchmark.sh b/databend/benchmark.sh
index a47a2f9a46..1e3398bfbd 100755
--- a/databend/benchmark.sh
+++ b/databend/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 # databend's first cold start (meta + query coming up together, opening
 # every object-store dir) regularly runs past the lib's 300s default;
diff --git a/datafusion-partitioned/benchmark.sh b/datafusion-partitioned/benchmark.sh
index 4e3c14373c..228e2176e4 100755
--- a/datafusion-partitioned/benchmark.sh
+++ b/datafusion-partitioned/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-partitioned"
 export BENCH_RESTARTABLE=no
 # skip concurrent_qps tests by default as datafusion-cli is configured
diff --git a/datafusion-vortex-partitioned/benchmark.sh b/datafusion-vortex-partitioned/benchmark.sh
index 8ec8a1ead6..c78ca0e002 100755
--- a/datafusion-vortex-partitioned/benchmark.sh
+++ b/datafusion-vortex-partitioned/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 # query_bench (the vortex driver) handles its own dataset download/conversion.
 export BENCH_DOWNLOAD_SCRIPT=""
 export BENCH_RESTARTABLE=no
diff --git a/datafusion-vortex/benchmark.sh b/datafusion-vortex/benchmark.sh
index e5ac9658f7..ae4a29cf69 100755
--- a/datafusion-vortex/benchmark.sh
+++ b/datafusion-vortex/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 # clickbench (the vortex driver) handles its own dataset download/conversion.
 export BENCH_DOWNLOAD_SCRIPT=""
 export BENCH_RESTARTABLE=no
diff --git a/datafusion/benchmark.sh b/datafusion/benchmark.sh
index 997d89cc61..4128a24124 100755
--- a/datafusion/benchmark.sh
+++ b/datafusion/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_RESTARTABLE=no
 # skip concurrent_qps tests by default as datafusion-cli is configured
diff --git a/doris-parquet/benchmark.sh b/doris-parquet/benchmark.sh
index 12a00b626c..d5cf12db78 100755
--- a/doris-parquet/benchmark.sh
+++ b/doris-parquet/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-partitioned"
 exec ../lib/benchmark-common.sh
diff --git a/doris/benchmark.sh b/doris/benchmark.sh
index 12a00b626c..d5cf12db78 100755
--- a/doris/benchmark.sh
+++ b/doris/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-partitioned"
 exec ../lib/benchmark-common.sh
diff --git a/drill/benchmark.sh b/drill/benchmark.sh
index bad104cedb..fb3b4d1318 100755
--- a/drill/benchmark.sh
+++ b/drill/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/druid/benchmark.sh b/druid/benchmark.sh
index 2a3572433c..f4a166051c 100755
--- a/druid/benchmark.sh
+++ b/druid/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 # Druid degrades after some queries; the shared driver restarts between
 # queries when restartable=yes (matches the original `pkill -f historical`
diff --git a/duckdb-dataframe/benchmark.sh b/duckdb-dataframe/benchmark.sh
index 4148581913..0a7321ea9c 100755
--- a/duckdb-dataframe/benchmark.sh
+++ b/duckdb-dataframe/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_DURABLE=no
 # Skip the pre-snapshot ./stop+./start cycle: the loaded
diff --git a/duckdb-datalake-partitioned/benchmark.sh b/duckdb-datalake-partitioned/benchmark.sh
index a5587169d1..d82911fa20 100755
--- a/duckdb-datalake-partitioned/benchmark.sh
+++ b/duckdb-datalake-partitioned/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 # Data is read directly from S3, no local download.
 export BENCH_DOWNLOAD_SCRIPT=""
 export BENCH_RESTARTABLE=no
diff --git a/duckdb-datalake/benchmark.sh b/duckdb-datalake/benchmark.sh
index a5587169d1..d82911fa20 100755
--- a/duckdb-datalake/benchmark.sh
+++ b/duckdb-datalake/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 # Data is read directly from S3, no local download.
 export BENCH_DOWNLOAD_SCRIPT=""
 export BENCH_RESTARTABLE=no
diff --git a/duckdb-memory/benchmark.sh b/duckdb-memory/benchmark.sh
index 6bf667e4f0..1fa99f381e 100755
--- a/duckdb-memory/benchmark.sh
+++ b/duckdb-memory/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_DURABLE=no
 exec ../lib/benchmark-common.sh
diff --git a/duckdb-parquet-partitioned/benchmark.sh b/duckdb-parquet-partitioned/benchmark.sh
index fea177f34d..b7ae16ae72 100755
--- a/duckdb-parquet-partitioned/benchmark.sh
+++ b/duckdb-parquet-partitioned/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-partitioned"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/duckdb-parquet/benchmark.sh b/duckdb-parquet/benchmark.sh
index bad104cedb..fb3b4d1318 100755
--- a/duckdb-parquet/benchmark.sh
+++ b/duckdb-parquet/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/duckdb-vortex-partitioned/benchmark.sh b/duckdb-vortex-partitioned/benchmark.sh
index fea177f34d..b7ae16ae72 100755
--- a/duckdb-vortex-partitioned/benchmark.sh
+++ b/duckdb-vortex-partitioned/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-partitioned"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/duckdb-vortex/benchmark.sh b/duckdb-vortex/benchmark.sh
index bad104cedb..fb3b4d1318 100755
--- a/duckdb-vortex/benchmark.sh
+++ b/duckdb-vortex/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/duckdb/benchmark.sh b/duckdb/benchmark.sh
index bad104cedb..fb3b4d1318 100755
--- a/duckdb/benchmark.sh
+++ b/duckdb/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/elasticsearch/benchmark.sh b/elasticsearch/benchmark.sh
index 118dd1e997..5aee2dc790 100755
--- a/elasticsearch/benchmark.sh
+++ b/elasticsearch/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 # Source data is gzipped NDJSON, fetched directly inside ./load.
 export BENCH_DOWNLOAD_SCRIPT=""
 exec ../lib/benchmark-common.sh
diff --git a/eventql/benchmark.sh b/eventql/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/eventql/benchmark.sh
+++ b/eventql/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/firebolt-parquet-partitioned/benchmark.sh b/firebolt-parquet-partitioned/benchmark.sh
index dbb9072c56..2b7c4464fe 100755
--- a/firebolt-parquet-partitioned/benchmark.sh
+++ b/firebolt-parquet-partitioned/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-partitioned"
 export BENCH_DURABLE=no
 export BENCH_RESTARTABLE=no
diff --git a/firebolt-parquet/benchmark.sh b/firebolt-parquet/benchmark.sh
index 3a332d30db..8fee8af389 100755
--- a/firebolt-parquet/benchmark.sh
+++ b/firebolt-parquet/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_DURABLE=no
 export BENCH_RESTARTABLE=no
diff --git a/firebolt/benchmark.sh b/firebolt/benchmark.sh
index fc4bacc8f3..3db3216601 100755
--- a/firebolt/benchmark.sh
+++ b/firebolt/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_RESTARTABLE=no
 exec ../lib/benchmark-common.sh
diff --git a/gendb/benchmark.sh b/gendb/benchmark.sh
index c5b5c24fd7..7128ac17b6 100755
--- a/gendb/benchmark.sh
+++ b/gendb/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 #
 # GenDB's pipeline produces one specialized C++ binary per query (the .cpp
 # files we ship in generated/ were synthesized by a multi-agent LLM pipeline
diff --git a/gizmosql/benchmark.sh b/gizmosql/benchmark.sh
index ddd7f28318..a33527ee30 100755
--- a/gizmosql/benchmark.sh
+++ b/gizmosql/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 exec ../lib/benchmark-common.sh
diff --git a/glaredb-partitioned/benchmark.sh b/glaredb-partitioned/benchmark.sh
index fea177f34d..b7ae16ae72 100755
--- a/glaredb-partitioned/benchmark.sh
+++ b/glaredb-partitioned/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-partitioned"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/glaredb/benchmark.sh b/glaredb/benchmark.sh
index bad104cedb..fb3b4d1318 100755
--- a/glaredb/benchmark.sh
+++ b/glaredb/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/greengage/benchmark.sh b/greengage/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/greengage/benchmark.sh
+++ b/greengage/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/greenplum/benchmark.sh b/greenplum/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/greenplum/benchmark.sh
+++ b/greenplum/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/heavyai/benchmark.sh b/heavyai/benchmark.sh
index 24602aa838..8426f806c5 100755
--- a/heavyai/benchmark.sh
+++ b/heavyai/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-csv"
 # omnisci/core-os-cpu's first cold start runs schema migrations, opens
 # its catalog, and binds Thrift ports; 600 s wasn't enough on the first
diff --git a/hive/benchmark.sh b/hive/benchmark.sh
index 0fbc3b8465..e16fbff562 100755
--- a/hive/benchmark.sh
+++ b/hive/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 # Embedded Derby metastore lives in the container's writable layer;
 # the cold-cycle docker rm + docker run in ./start wipes it. ./load
diff --git a/hyper-parquet/benchmark.sh b/hyper-parquet/benchmark.sh
index 12a00b626c..d5cf12db78 100755
--- a/hyper-parquet/benchmark.sh
+++ b/hyper-parquet/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-partitioned"
 exec ../lib/benchmark-common.sh
diff --git a/hyper/benchmark.sh b/hyper/benchmark.sh
index 7d4ef95c38..f61f185e49 100755
--- a/hyper/benchmark.sh
+++ b/hyper/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-csv"
 exec ../lib/benchmark-common.sh
diff --git a/impala/benchmark.sh b/impala/benchmark.sh
index fd9ea899f6..fa3a0559bc 100755
--- a/impala/benchmark.sh
+++ b/impala/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 #
 # First-cold Impala start has to bootstrap the Hive metastore schema,
 # wait for catalogd to register with statestored, and only then does
diff --git a/infobright/benchmark.sh b/infobright/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/infobright/benchmark.sh
+++ b/infobright/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/kdb/benchmark.sh b/kdb/benchmark.sh
index bad104cedb..fb3b4d1318 100755
--- a/kdb/benchmark.sh
+++ b/kdb/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/kinetica/benchmark.sh b/kinetica/benchmark.sh
index 6324349f94..ce146975c8 100755
--- a/kinetica/benchmark.sh
+++ b/kinetica/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 # kinetica downloads hits.tsv.gz directly inside ./load (Kinetica wants the
 # gzipped form), so no central download script is used.
 export BENCH_DOWNLOAD_SCRIPT=""
diff --git a/locustdb/benchmark.sh b/locustdb/benchmark.sh
index 7d4ef95c38..f61f185e49 100755
--- a/locustdb/benchmark.sh
+++ b/locustdb/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-csv"
 exec ../lib/benchmark-common.sh
diff --git a/mariadb-columnstore/benchmark.sh b/mariadb-columnstore/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/mariadb-columnstore/benchmark.sh
+++ b/mariadb-columnstore/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/mariadb-duckdb/benchmark.sh b/mariadb-duckdb/benchmark.sh
index 33f0e78378..124a77209b 100755
--- a/mariadb-duckdb/benchmark.sh
+++ b/mariadb-duckdb/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in ../lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/mariadb/benchmark.sh b/mariadb/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/mariadb/benchmark.sh
+++ b/mariadb/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/monetdb/benchmark.sh b/monetdb/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/monetdb/benchmark.sh
+++ b/monetdb/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/mongodb/benchmark.sh b/mongodb/benchmark.sh
index 10ccc1cc79..fee6836c31 100755
--- a/mongodb/benchmark.sh
+++ b/mongodb/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh. MongoDB uses
 # aggregation pipelines (queries.txt, EJSON one-per-line) rather than SQL.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 export BENCH_QUERIES_FILE="queries.txt"
diff --git a/mssql/benchmark.sh b/mssql/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/mssql/benchmark.sh
+++ b/mssql/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/mysql-myisam/benchmark.sh b/mysql-myisam/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/mysql-myisam/benchmark.sh
+++ b/mysql-myisam/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/mysql/benchmark.sh b/mysql/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/mysql/benchmark.sh
+++ b/mysql/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/octosql/benchmark.sh b/octosql/benchmark.sh
index bad104cedb..fb3b4d1318 100755
--- a/octosql/benchmark.sh
+++ b/octosql/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/opengpdb/benchmark.sh b/opengpdb/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/opengpdb/benchmark.sh
+++ b/opengpdb/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/opteryx/benchmark.sh b/opteryx/benchmark.sh
index fea177f34d..b7ae16ae72 100755
--- a/opteryx/benchmark.sh
+++ b/opteryx/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-partitioned"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/oxla/benchmark.sh b/oxla/benchmark.sh
index 7d4ef95c38..f61f185e49 100755
--- a/oxla/benchmark.sh
+++ b/oxla/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-csv"
 exec ../lib/benchmark-common.sh
diff --git a/pandas/benchmark.sh b/pandas/benchmark.sh
index 1369feb230..09da9ac59a 100755
--- a/pandas/benchmark.sh
+++ b/pandas/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_DURABLE=no
 # pandas runs Python expressions directly (server eval()s them).
diff --git a/paradedb-partitioned/benchmark.sh b/paradedb-partitioned/benchmark.sh
index 12a00b626c..d5cf12db78 100755
--- a/paradedb-partitioned/benchmark.sh
+++ b/paradedb-partitioned/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-partitioned"
 exec ../lib/benchmark-common.sh
diff --git a/paradedb/benchmark.sh b/paradedb/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/paradedb/benchmark.sh
+++ b/paradedb/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/parseable/benchmark.sh b/parseable/benchmark.sh
index 4239630008..535701276c 100755
--- a/parseable/benchmark.sh
+++ b/parseable/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-json"
 exec ../lib/benchmark-common.sh
diff --git a/pg_clickhouse/benchmark.sh b/pg_clickhouse/benchmark.sh
index 12a00b626c..d5cf12db78 100755
--- a/pg_clickhouse/benchmark.sh
+++ b/pg_clickhouse/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-partitioned"
 exec ../lib/benchmark-common.sh
diff --git a/pg_deltax/benchmark.sh b/pg_deltax/benchmark.sh
index 12a00b626c..d5cf12db78 100755
--- a/pg_deltax/benchmark.sh
+++ b/pg_deltax/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-partitioned"
 exec ../lib/benchmark-common.sh
diff --git a/pg_duckdb-indexed/benchmark.sh b/pg_duckdb-indexed/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/pg_duckdb-indexed/benchmark.sh
+++ b/pg_duckdb-indexed/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/pg_duckdb-motherduck/benchmark.sh b/pg_duckdb-motherduck/benchmark.sh
index b2dd93459a..3b046ae9f9 100755
--- a/pg_duckdb-motherduck/benchmark.sh
+++ b/pg_duckdb-motherduck/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 # Empty BENCH_DOWNLOAD_SCRIPT: the data lives in MotherDuck cloud (the
 # load script CTAS'es directly from S3 inside MotherDuck), nothing to
 # fetch locally.
diff --git a/pg_duckdb-parquet/benchmark.sh b/pg_duckdb-parquet/benchmark.sh
index 5e78e0128a..51e0cdf61e 100755
--- a/pg_duckdb-parquet/benchmark.sh
+++ b/pg_duckdb-parquet/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 # Empty BENCH_DOWNLOAD_SCRIPT: install fetches hits.parquet itself,
 # because the container needs the file bind-mounted at start time
 # (before lib's bench_download step runs).
diff --git a/pg_duckdb/benchmark.sh b/pg_duckdb/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/pg_duckdb/benchmark.sh
+++ b/pg_duckdb/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/pg_ducklake/benchmark.sh b/pg_ducklake/benchmark.sh
index ddd7f28318..a33527ee30 100755
--- a/pg_ducklake/benchmark.sh
+++ b/pg_ducklake/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 exec ../lib/benchmark-common.sh
diff --git a/pg_mooncake/benchmark.sh b/pg_mooncake/benchmark.sh
index ddd7f28318..a33527ee30 100755
--- a/pg_mooncake/benchmark.sh
+++ b/pg_mooncake/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 exec ../lib/benchmark-common.sh
diff --git a/pgpro_tam/benchmark.sh b/pgpro_tam/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/pgpro_tam/benchmark.sh
+++ b/pgpro_tam/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/pinot/benchmark.sh b/pinot/benchmark.sh
index cea0b3513c..e5db3195ea 100755
--- a/pinot/benchmark.sh
+++ b/pinot/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 # Pinot's quickstart starts a controller, broker, server and a Zookeeper
 # inside one JVM and takes longer than the lib's 300 s default to be
diff --git a/polars-dataframe/benchmark.sh b/polars-dataframe/benchmark.sh
index a5a40fb651..e798b8a668 100755
--- a/polars-dataframe/benchmark.sh
+++ b/polars-dataframe/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_DURABLE=no
 # polars runs Python expressions directly (server eval()s them).
diff --git a/polars/benchmark.sh b/polars/benchmark.sh
index 4148581913..0a7321ea9c 100755
--- a/polars/benchmark.sh
+++ b/polars/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_DURABLE=no
 # Skip the pre-snapshot ./stop+./start cycle: the loaded
diff --git a/postgresql-indexed/benchmark.sh b/postgresql-indexed/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/postgresql-indexed/benchmark.sh
+++ b/postgresql-indexed/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/postgresql-orioledb/benchmark.sh b/postgresql-orioledb/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/postgresql-orioledb/benchmark.sh
+++ b/postgresql-orioledb/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/postgresql/benchmark.sh b/postgresql/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/postgresql/benchmark.sh
+++ b/postgresql/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/presto-datalake-partitioned/benchmark.sh b/presto-datalake-partitioned/benchmark.sh
index 4937ea0457..bf1aa44e29 100755
--- a/presto-datalake-partitioned/benchmark.sh
+++ b/presto-datalake-partitioned/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 # Datalake variant: Parquet is read directly from public S3, no download.
 export BENCH_DOWNLOAD_SCRIPT=""
 exec ../lib/benchmark-common.sh
diff --git a/presto-datalake/benchmark.sh b/presto-datalake/benchmark.sh
index 4937ea0457..bf1aa44e29 100755
--- a/presto-datalake/benchmark.sh
+++ b/presto-datalake/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 # Datalake variant: Parquet is read directly from public S3, no download.
 export BENCH_DOWNLOAD_SCRIPT=""
 exec ../lib/benchmark-common.sh
diff --git a/presto-partitioned/benchmark.sh b/presto-partitioned/benchmark.sh
index 12a00b626c..d5cf12db78 100755
--- a/presto-partitioned/benchmark.sh
+++ b/presto-partitioned/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-partitioned"
 exec ../lib/benchmark-common.sh
diff --git a/presto/benchmark.sh b/presto/benchmark.sh
index ddd7f28318..a33527ee30 100755
--- a/presto/benchmark.sh
+++ b/presto/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 exec ../lib/benchmark-common.sh
diff --git a/questdb/benchmark.sh b/questdb/benchmark.sh
index 7d4ef95c38..f61f185e49 100755
--- a/questdb/benchmark.sh
+++ b/questdb/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-csv"
 exec ../lib/benchmark-common.sh
diff --git a/quickwit/benchmark.sh b/quickwit/benchmark.sh
index 5d5d7e2479..2e3f2c3fd7 100755
--- a/quickwit/benchmark.sh
+++ b/quickwit/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 # Quickwit takes Elasticsearch-format JSON queries; the load script fetches
 # hits.json.gz directly so no shared download-hits-* script applies.
 export BENCH_DOWNLOAD_SCRIPT=""
diff --git a/sail-partitioned/benchmark.sh b/sail-partitioned/benchmark.sh
index fea177f34d..b7ae16ae72 100755
--- a/sail-partitioned/benchmark.sh
+++ b/sail-partitioned/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-partitioned"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/sail/benchmark.sh b/sail/benchmark.sh
index bad104cedb..fb3b4d1318 100755
--- a/sail/benchmark.sh
+++ b/sail/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/siglens/benchmark.sh b/siglens/benchmark.sh
index 134d3b88d8..2288da297c 100755
--- a/siglens/benchmark.sh
+++ b/siglens/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 # siglens ingests its own gzipped NDJSON; ./load fetches it directly.
 export BENCH_DOWNLOAD_SCRIPT=""
 # queries are SPL/Splunk QL, not SQL.
diff --git a/sirius/benchmark.sh b/sirius/benchmark.sh
index 9432ba26d9..c4bbbd067b 100755
--- a/sirius/benchmark.sh
+++ b/sirius/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_DURABLE=no
 # sirius's server.py initializes CUDA / cuDF on startup which can take
diff --git a/spark-auron/benchmark.sh b/spark-auron/benchmark.sh
index bad104cedb..fb3b4d1318 100755
--- a/spark-auron/benchmark.sh
+++ b/spark-auron/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/spark-comet/benchmark.sh b/spark-comet/benchmark.sh
index bad104cedb..fb3b4d1318 100755
--- a/spark-comet/benchmark.sh
+++ b/spark-comet/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/spark-gluten/benchmark.sh b/spark-gluten/benchmark.sh
index bad104cedb..fb3b4d1318 100755
--- a/spark-gluten/benchmark.sh
+++ b/spark-gluten/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/spark-velox/benchmark.sh b/spark-velox/benchmark.sh
index bad104cedb..fb3b4d1318 100755
--- a/spark-velox/benchmark.sh
+++ b/spark-velox/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/spark/benchmark.sh b/spark/benchmark.sh
index 7a306e5f9f..598f8a07ad 100755
--- a/spark/benchmark.sh
+++ b/spark/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 # Spark runs in-process per query — restart between queries is meaningless
 # (and would re-download nothing). Skip restart.
diff --git a/sqlite/benchmark.sh b/sqlite/benchmark.sh
index 1b692a2155..75c3806085 100755
--- a/sqlite/benchmark.sh
+++ b/sqlite/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-csv"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/starrocks/benchmark.sh b/starrocks/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/starrocks/benchmark.sh
+++ b/starrocks/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/tidb/benchmark.sh b/tidb/benchmark.sh
index 8d9d6fb616..81970cac48 100755
--- a/tidb/benchmark.sh
+++ b/tidb/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 # TiDB Lightning loads from <db>.<table>.csv files; we use the CSV download.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-csv"
 # Skip the pre-snapshot ./stop+./start cycle: the loaded
diff --git a/timescaledb-no-columnstore/benchmark.sh b/timescaledb-no-columnstore/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/timescaledb-no-columnstore/benchmark.sh
+++ b/timescaledb-no-columnstore/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/timescaledb/benchmark.sh b/timescaledb/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/timescaledb/benchmark.sh
+++ b/timescaledb/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/trino-datalake-partitioned/benchmark.sh b/trino-datalake-partitioned/benchmark.sh
index dbe8f14075..174bc76ff7 100755
--- a/trino-datalake-partitioned/benchmark.sh
+++ b/trino-datalake-partitioned/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 # Datalake variant: Parquet is read directly from public S3, no download.
 export BENCH_DOWNLOAD_SCRIPT=""
 # Trino bootstrap on a cold sysdisk pushes past the 300s default.
diff --git a/trino-datalake/benchmark.sh b/trino-datalake/benchmark.sh
index dbe8f14075..174bc76ff7 100755
--- a/trino-datalake/benchmark.sh
+++ b/trino-datalake/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 # Datalake variant: Parquet is read directly from public S3, no download.
 export BENCH_DOWNLOAD_SCRIPT=""
 # Trino bootstrap on a cold sysdisk pushes past the 300s default.
diff --git a/trino-partitioned/benchmark.sh b/trino-partitioned/benchmark.sh
index 12a00b626c..d5cf12db78 100755
--- a/trino-partitioned/benchmark.sh
+++ b/trino-partitioned/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-partitioned"
 exec ../lib/benchmark-common.sh
diff --git a/trino/benchmark.sh b/trino/benchmark.sh
index ddd7f28318..a33527ee30 100755
--- a/trino/benchmark.sh
+++ b/trino/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-single"
 exec ../lib/benchmark-common.sh
diff --git a/turso/benchmark.sh b/turso/benchmark.sh
index 1b692a2155..75c3806085 100755
--- a/turso/benchmark.sh
+++ b/turso/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-csv"
 export BENCH_RESTARTABLE=no
 # Single-process engine: each query forks a fresh full-machine process with no
diff --git a/umbra/benchmark.sh b/umbra/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/umbra/benchmark.sh
+++ b/umbra/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/ursa/benchmark.sh b/ursa/benchmark.sh
index 12a00b626c..d5cf12db78 100755
--- a/ursa/benchmark.sh
+++ b/ursa/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-parquet-partitioned"
 exec ../lib/benchmark-common.sh
diff --git a/velodb/benchmark.sh b/velodb/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/velodb/benchmark.sh
+++ b/velodb/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/vertica/benchmark.sh b/vertica/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/vertica/benchmark.sh
+++ b/vertica/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/victorialogs/benchmark.sh b/victorialogs/benchmark.sh
index 26051b2260..31a8f369e5 100755
--- a/victorialogs/benchmark.sh
+++ b/victorialogs/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-json"
 # queries are LogsQL, not SQL.
 export BENCH_QUERIES_FILE="queries.logsql"
diff --git a/warehousepg/benchmark.sh b/warehousepg/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/warehousepg/benchmark.sh
+++ b/warehousepg/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh
diff --git a/ydb/benchmark.sh b/ydb/benchmark.sh
index 836598f4f8..31fa4394ba 100755
--- a/ydb/benchmark.sh
+++ b/ydb/benchmark.sh
@@ -1,5 +1,4 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 # YDB downloads CSV directly inside ./load (the ydb CLI imports from CSV).
 export BENCH_DOWNLOAD_SCRIPT=""
 # YDB has no benefit from server restart — it's a multi-node distributed
diff --git a/yugabytedb/benchmark.sh b/yugabytedb/benchmark.sh
index a5c2d951ec..124a77209b 100755
--- a/yugabytedb/benchmark.sh
+++ b/yugabytedb/benchmark.sh
@@ -1,4 +1,3 @@
 #!/bin/bash
-# Thin shim — actual flow is in lib/benchmark-common.sh.
 export BENCH_DOWNLOAD_SCRIPT="download-hits-tsv"
 exec ../lib/benchmark-common.sh