diff --git a/modules/supabase/module.json b/modules/supabase/module.json
index 7d613cd..c578306 100644
--- a/modules/supabase/module.json
+++ b/modules/supabase/module.json
@@ -37,7 +37,10 @@
     "service-role-key": "${dir:state}/service-role-key.secret",
     "dashboard-user": "${dir:state}/dashboard-user.secret",
     "dashboard": "${dir:state}/dashboard.secret",
-    "logflare": "${dir:state}/logflare.secret",
+    "logflare": {
+      "path": "${dir:state}/logflare.secret",
+      "taken": "applied"
+    },
     "pooler-vault": "${dir:state}/pooler-vault.secret",
     "key-base-a": "${dir:state}/key-base-a.secret",
     "key-base-b": "${dir:state}/key-base-b.secret",
@@ -130,7 +133,7 @@
       "type": "file",
       "path": "${dir:state}/vector.yml",
       "mode": "0600",
-      "content": "# Written by the mesh (modules/supabase).\napi:\n  enabled: true\n  address: 0.0.0.0:9001\n\nsources:\n  docker_host:\n    type: docker_logs\n    include_containers:\n      - supabase-kong\n      - supabase-auth\n      - supabase-rest\n      - realtime-dev.supabase-realtime\n      - supabase-storage\n      - supabase-edge-functions\n      - supabase-db\n\ntransforms:\n  project_logs:\n    type: remap\n    inputs:\n      - docker_host\n    source: |-\n      .project = \"default\"\n      .event_message = del(.message)\n      .appname = del(.container_name)\n      del(.container_created_at)\n      del(.container_id)\n      del(.source_type)\n      del(.stream)\n      del(.label)\n      del(.image)\n      del(.host)\n      del(.stream)\n  router:\n    type: route\n    inputs:\n      - project_logs\n    route:\n      kong: '.appname == \"supabase-kong\"'\n      auth: '.appname == \"supabase-auth\"'\n      rest: '.appname == \"supabase-rest\"'\n      realtime: '.appname == \"realtime-dev.supabase-realtime\"'\n      storage: '.appname == \"supabase-storage\"'\n      functions: '.appname == \"supabase-edge-functions\"'\n      db: '.appname == \"supabase-db\"'\n  # Ignores non nginx errors since they are related with kong booting up\n  kong_logs:\n    type: remap\n    inputs:\n      - router.kong\n    source: |-\n      req, err = parse_nginx_log(.event_message, \"combined\")\n      if err == null {\n          .timestamp = req.timestamp\n          .metadata.request.headers.referer = req.referer\n          .metadata.request.headers.user_agent = req.agent\n          .metadata.request.headers.cf_connecting_ip = req.client\n          .metadata.request.method = req.method\n          .metadata.request.path = req.path\n          .metadata.request.protocol = req.protocol\n          .metadata.response.status_code = req.status\n      }\n      if err != null {\n        abort\n      }\n  # Ignores non nginx errors since they are related with kong booting up\n  kong_err:\n    type: remap\n    inputs:\n      - router.kong\n    source: |-\n      .metadata.request.method = \"GET\"\n      .metadata.response.status_code = 200\n      parsed, err = parse_nginx_log(.event_message, \"error\")\n      if err == null {\n          .timestamp = parsed.timestamp\n          .severity = parsed.severity\n          .metadata.request.host = parsed.host\n          .metadata.request.headers.cf_connecting_ip = parsed.client\n          url, err = split(parsed.request, \" \")\n          if err == null {\n              .metadata.request.method = url[0]\n              .metadata.request.path = url[1]\n              .metadata.request.protocol = url[2]\n          }\n      }\n      if err != null {\n        abort\n      }\n  # Gotrue logs are structured json strings which frontend parses directly. But we keep metadata for consistency.\n  auth_logs:\n    type: remap\n    inputs:\n      - router.auth\n    source: |-\n      parsed, err = parse_json(.event_message)\n      if err == null {\n          .metadata.timestamp = parsed.time\n          .metadata = merge!(.metadata, parsed)\n      }\n  # PostgREST logs are structured so we separate timestamp from message using regex\n  rest_logs:\n    type: remap\n    inputs:\n      - router.rest\n    source: |-\n      parsed, err = parse_regex(.event_message, r'^(?P<time>.*): (?P<msg>.*)$')\n      if err == null {\n          .event_message = parsed.msg\n          .timestamp = to_timestamp!(parsed.time)\n          .metadata.host = .project\n      }\n  # Realtime logs are structured so we parse the severity level using regex (ignore time because it has no date)\n  realtime_logs:\n    type: remap\n    inputs:\n      - router.realtime\n    source: |-\n      .metadata.project = del(.project)\n      .metadata.external_id = .metadata.project\n      parsed, err = parse_regex(.event_message, r'^(?P<time>\\d+:\\d+:\\d+\\.\\d+) \\[(?P<level>\\w+)\\] (?P<msg>.*)$')\n      if err == null {\n          .event_message = parsed.msg\n          .metadata.level = parsed.level\n      }\n  # Storage logs may contain json objects so we parse them for completeness\n  storage_logs:\n    type: remap\n    inputs:\n      - router.storage\n    source: |-\n      .metadata.project = del(.project)\n      .metadata.tenantId = .metadata.project\n      parsed, err = parse_json(.event_message)\n      if err == null {\n          .event_message = parsed.msg\n          .metadata.level = parsed.level\n          .metadata.timestamp = parsed.time\n          .metadata.context[0].host = parsed.hostname\n          .metadata.context[0].pid = parsed.pid\n      }\n  # Postgres logs some messages to stderr which we map to warning severity level\n  db_logs:\n    type: remap\n    inputs:\n      - router.db\n    source: |-\n      .metadata.host = \"db-default\"\n      .metadata.parsed.timestamp = .timestamp\n\n      parsed, err = parse_regex(.event_message, r'.*(?P<level>INFO|NOTICE|WARNING|ERROR|LOG|FATAL|PANIC?):.*', numeric_groups: true)\n\n      if err != null || parsed == null {\n        .metadata.parsed.error_severity = \"info\"\n      }\n      if parsed != null {\n       .metadata.parsed.error_severity = parsed.level\n      }\n      if .metadata.parsed.error_severity == \"info\" {\n          .metadata.parsed.error_severity = \"log\"\n      }\n      .metadata.parsed.error_severity = upcase!(.metadata.parsed.error_severity)\n\nsinks:\n  logflare_auth:\n    type: 'http'\n    inputs:\n      - auth_logs\n    encoding:\n      codec: 'json'\n    method: 'post'\n    request:\n      retry_max_duration_secs: 10\n    uri: 'http://supabase-analytics:4000/api/logs?source_name=gotrue.logs.prod&api_key=${secret:logflare}'\n  logflare_realtime:\n    type: 'http'\n    inputs:\n      - realtime_logs\n    encoding:\n      codec: 'json'\n    method: 'post'\n    request:\n      retry_max_duration_secs: 10\n    uri: 'http://supabase-analytics:4000/api/logs?source_name=realtime.logs.prod&api_key=${secret:logflare}'\n  logflare_rest:\n    type: 'http'\n    inputs:\n      - rest_logs\n    encoding:\n      codec: 'json'\n    method: 'post'\n    request:\n      retry_max_duration_secs: 10\n    uri: 'http://supabase-analytics:4000/api/logs?source_name=postgREST.logs.prod&api_key=${secret:logflare}'\n  logflare_db:\n    type: 'http'\n    inputs:\n      - db_logs\n    encoding:\n      codec: 'json'\n    method: 'post'\n    request:\n      retry_max_duration_secs: 10\n    # We must route the sink through kong because ingesting logs before logflare is fully initialised will\n    # lead to broken queries from studio. This works by the assumption that containers are started in the\n    # following order: vector > db > logflare > kong\n    uri: 'http://supabase-kong:8000/analytics/v1/api/logs?source_name=postgres.logs&api_key=${secret:logflare}'\n  logflare_functions:\n    type: 'http'\n    inputs:\n      - router.functions\n    encoding:\n      codec: 'json'\n    method: 'post'\n    request:\n      retry_max_duration_secs: 10\n    uri: 'http://supabase-analytics:4000/api/logs?source_name=deno-relay-logs&api_key=${secret:logflare}'\n  logflare_storage:\n    type: 'http'\n    inputs:\n      - storage_logs\n    encoding:\n      codec: 'json'\n    method: 'post'\n    request:\n      retry_max_duration_secs: 10\n    uri: 'http://supabase-analytics:4000/api/logs?source_name=storage.logs.prod.2&api_key=${secret:logflare}'\n  logflare_kong:\n    type: 'http'\n    inputs:\n      - kong_logs\n      - kong_err\n    encoding:\n      codec: 'json'\n    method: 'post'\n    request:\n      retry_max_duration_secs: 10\n    uri: 'http://supabase-analytics:4000/api/logs?source_name=cloudflare.logs.prod&api_key=${secret:logflare}'\n"
+      "content": "# Written by the mesh (modules/supabase).\n#\n# Every sink hands logflare its API key in the x-api-key header, never as an api_key query parameter:\n# logflare prints a failed request's whole URL, query string included, in its error report, so a\n# key in the URL is a key in its log (novox/hq issue 268). supabase-analytics refuses to start\n# while this file still puts the key in a URL.\napi:\n  enabled: true\n  address: 0.0.0.0:9001\n\nsources:\n  docker_host:\n    type: docker_logs\n    include_containers:\n      - supabase-kong\n      - supabase-auth\n      - supabase-rest\n      - realtime-dev.supabase-realtime\n      - supabase-storage\n      - supabase-edge-functions\n      - supabase-db\n\ntransforms:\n  project_logs:\n    type: remap\n    inputs:\n      - docker_host\n    source: |-\n      .project = \"default\"\n      .event_message = del(.message)\n      .appname = del(.container_name)\n      del(.container_created_at)\n      del(.container_id)\n      del(.source_type)\n      del(.stream)\n      del(.label)\n      del(.image)\n      del(.host)\n      del(.stream)\n  router:\n    type: route\n    inputs:\n      - project_logs\n    route:\n      kong: '.appname == \"supabase-kong\"'\n      auth: '.appname == \"supabase-auth\"'\n      rest: '.appname == \"supabase-rest\"'\n      realtime: '.appname == \"realtime-dev.supabase-realtime\"'\n      storage: '.appname == \"supabase-storage\"'\n      functions: '.appname == \"supabase-edge-functions\"'\n      db: '.appname == \"supabase-db\"'\n  # Ignores non nginx errors since they are related with kong booting up\n  kong_logs:\n    type: remap\n    inputs:\n      - router.kong\n    source: |-\n      req, err = parse_nginx_log(.event_message, \"combined\")\n      if err == null {\n          .timestamp = req.timestamp\n          .metadata.request.headers.referer = req.referer\n          .metadata.request.headers.user_agent = req.agent\n          .metadata.request.headers.cf_connecting_ip = req.client\n          .metadata.request.method = req.method\n          .metadata.request.path = req.path\n          .metadata.request.protocol = req.protocol\n          .metadata.response.status_code = req.status\n      }\n      if err != null {\n        abort\n      }\n  # Ignores non nginx errors since they are related with kong booting up\n  kong_err:\n    type: remap\n    inputs:\n      - router.kong\n    source: |-\n      .metadata.request.method = \"GET\"\n      .metadata.response.status_code = 200\n      parsed, err = parse_nginx_log(.event_message, \"error\")\n      if err == null {\n          .timestamp = parsed.timestamp\n          .severity = parsed.severity\n          .metadata.request.host = parsed.host\n          .metadata.request.headers.cf_connecting_ip = parsed.client\n          url, err = split(parsed.request, \" \")\n          if err == null {\n              .metadata.request.method = url[0]\n              .metadata.request.path = url[1]\n              .metadata.request.protocol = url[2]\n          }\n      }\n      if err != null {\n        abort\n      }\n  # Gotrue logs are structured json strings which frontend parses directly. But we keep metadata for consistency.\n  auth_logs:\n    type: remap\n    inputs:\n      - router.auth\n    source: |-\n      parsed, err = parse_json(.event_message)\n      if err == null {\n          .metadata.timestamp = parsed.time\n          .metadata = merge!(.metadata, parsed)\n      }\n  # PostgREST logs are structured so we separate timestamp from message using regex\n  rest_logs:\n    type: remap\n    inputs:\n      - router.rest\n    source: |-\n      parsed, err = parse_regex(.event_message, r'^(?P<time>.*): (?P<msg>.*)$')\n      if err == null {\n          .event_message = parsed.msg\n          .timestamp = to_timestamp!(parsed.time)\n          .metadata.host = .project\n      }\n  # Realtime logs are structured so we parse the severity level using regex (ignore time because it has no date)\n  realtime_logs:\n    type: remap\n    inputs:\n      - router.realtime\n    source: |-\n      .metadata.project = del(.project)\n      .metadata.external_id = .metadata.project\n      parsed, err = parse_regex(.event_message, r'^(?P<time>\\d+:\\d+:\\d+\\.\\d+) \\[(?P<level>\\w+)\\] (?P<msg>.*)$')\n      if err == null {\n          .event_message = parsed.msg\n          .metadata.level = parsed.level\n      }\n  # Storage logs may contain json objects so we parse them for completeness\n  storage_logs:\n    type: remap\n    inputs:\n      - router.storage\n    source: |-\n      .metadata.project = del(.project)\n      .metadata.tenantId = .metadata.project\n      parsed, err = parse_json(.event_message)\n      if err == null {\n          .event_message = parsed.msg\n          .metadata.level = parsed.level\n          .metadata.timestamp = parsed.time\n          .metadata.context[0].host = parsed.hostname\n          .metadata.context[0].pid = parsed.pid\n      }\n  # Postgres logs some messages to stderr which we map to warning severity level\n  db_logs:\n    type: remap\n    inputs:\n      - router.db\n    source: |-\n      .metadata.host = \"db-default\"\n      .metadata.parsed.timestamp = .timestamp\n\n      parsed, err = parse_regex(.event_message, r'.*(?P<level>INFO|NOTICE|WARNING|ERROR|LOG|FATAL|PANIC?):.*', numeric_groups: true)\n\n      if err != null || parsed == null {\n        .metadata.parsed.error_severity = \"info\"\n      }\n      if parsed != null {\n       .metadata.parsed.error_severity = parsed.level\n      }\n      if .metadata.parsed.error_severity == \"info\" {\n          .metadata.parsed.error_severity = \"log\"\n      }\n      .metadata.parsed.error_severity = upcase!(.metadata.parsed.error_severity)\n\nsinks:\n  logflare_auth:\n    type: 'http'\n    inputs:\n      - auth_logs\n    encoding:\n      codec: 'json'\n    method: 'post'\n    request:\n      retry_max_duration_secs: 10\n      headers:\n        x-api-key: '${secret:logflare}'\n    uri: 'http://supabase-analytics:4000/api/logs?source_name=gotrue.logs.prod'\n  logflare_realtime:\n    type: 'http'\n    inputs:\n      - realtime_logs\n    encoding:\n      codec: 'json'\n    method: 'post'\n    request:\n      retry_max_duration_secs: 10\n      headers:\n        x-api-key: '${secret:logflare}'\n    uri: 'http://supabase-analytics:4000/api/logs?source_name=realtime.logs.prod'\n  logflare_rest:\n    type: 'http'\n    inputs:\n      - rest_logs\n    encoding:\n      codec: 'json'\n    method: 'post'\n    request:\n      retry_max_duration_secs: 10\n      headers:\n        x-api-key: '${secret:logflare}'\n    uri: 'http://supabase-analytics:4000/api/logs?source_name=postgREST.logs.prod'\n  logflare_db:\n    type: 'http'\n    inputs:\n      - db_logs\n    encoding:\n      codec: 'json'\n    method: 'post'\n    request:\n      retry_max_duration_secs: 10\n      headers:\n        x-api-key: '${secret:logflare}'\n    # We must route the sink through kong because ingesting logs before logflare is fully initialised will\n    # lead to broken queries from studio. This works by the assumption that containers are started in the\n    # following order: vector > db > logflare > kong\n    uri: 'http://supabase-kong:8000/analytics/v1/api/logs?source_name=postgres.logs'\n  logflare_functions:\n    type: 'http'\n    inputs:\n      - router.functions\n    encoding:\n      codec: 'json'\n    method: 'post'\n    request:\n      retry_max_duration_secs: 10\n      headers:\n        x-api-key: '${secret:logflare}'\n    uri: 'http://supabase-analytics:4000/api/logs?source_name=deno-relay-logs'\n  logflare_storage:\n    type: 'http'\n    inputs:\n      - storage_logs\n    encoding:\n      codec: 'json'\n    method: 'post'\n    request:\n      retry_max_duration_secs: 10\n      headers:\n        x-api-key: '${secret:logflare}'\n    uri: 'http://supabase-analytics:4000/api/logs?source_name=storage.logs.prod.2'\n  logflare_kong:\n    type: 'http'\n    inputs:\n      - kong_logs\n      - kong_err\n    encoding:\n      codec: 'json'\n    method: 'post'\n    request:\n      retry_max_duration_secs: 10\n      headers:\n        x-api-key: '${secret:logflare}'\n    uri: 'http://supabase-analytics:4000/api/logs?source_name=cloudflare.logs.prod'\n"
     },
     {
       "id": "pooler-conf",
@@ -237,6 +240,13 @@
       "mode": "0600",
       "content": "LOGFLARE_NODE_HOST=127.0.0.1\nDB_USERNAME=supabase_admin\nDB_DATABASE=_supabase\nDB_HOSTNAME=supabase-db\nDB_PORT=5432\nDB_PASSWORD=${secret:postgres}\nDB_SCHEMA=_analytics\nLOGFLARE_API_KEY=${secret:logflare}\nLOGFLARE_SINGLE_TENANT=true\nLOGFLARE_SUPABASE_MODE=true\nLOGFLARE_MIN_CLUSTER_SIZE=1\nPOSTGRES_BACKEND_URL=postgresql://supabase_admin:${secret:postgres}@supabase-db:5432/_supabase\nPOSTGRES_BACKEND_SCHEMA=_analytics\nLOGFLARE_FEATURE_FLAG_OVERRIDE=multibackend=true\n"
     },
+    {
+      "id": "analytics-start",
+      "type": "file",
+      "path": "${dir:state}/analytics-start.sh",
+      "mode": "0644",
+      "content": "#!/bin/sh\n# Generated by the mesh. Do not edit: module supabase writes this file and replaces it at every push.\n#\n# logflare 1.4.0 prints the whole URL of every request that fails, query string included, in the\n# error report Plug.Cowboy writes (\"Request: POST /api/logs?source_name=...&api_key=...\"). That is\n# an error, so no log level hides it while errors are logged at all (novox/hq issue 268). An API key\n# passed in the URL is therefore in this container's log. logflare reads the key from the\n# x-api-key header as well, and the mesh's caller, vector, sends it there; this refuses to start\n# logflare while the vector config it is given still passes the key in a URL: an analytics that\n# does not start says why here, and one that leaks says nothing.\nset -e\nconf=/run/logflare/vector.yml\nif [ ! -r \"$conf\" ]; then\n  echo \"analytics: $conf is not there to check, so whether the API key travels in a URL is unknown; not starting (novox/hq issue 268)\" >&2\n  exit 1\nfi\nif grep -v '^[[:space:]]*#' \"$conf\" | grep -q 'api_key='; then\n  echo \"analytics: $conf passes the API key in a URL, and logflare prints a failed request's URL; not starting (novox/hq issue 268)\" >&2\n  exit 1\nfi\ncd /opt/app/rel/logflare/bin\nexec sh run.sh\n"
+    },
     {
       "id": "studio-env",
       "type": "file",
@@ -346,7 +356,19 @@
       "env-file": [
         "${dir:state}/analytics-env.env"
       ],
-      "secrets-in-environment": "Logflare reads its database password and API key from the environment only"
+      "volumes": [
+        "${dir:state}/analytics-start.sh:/run/logflare/start.sh:ro",
+        "${dir:state}/vector.yml:/run/logflare/vector.yml:ro"
+      ],
+      "args": [
+        "sh",
+        "/run/logflare/start.sh"
+      ],
+      "restart-on": [
+        "analytics-start",
+        "vector-conf"
+      ],
+      "secrets-in-environment": "Logflare reads its database password and API key from the environment only. The API key is never in a URL the mesh writes: logflare prints a failed request's URL, so vector sends the key in the x-api-key header, and the start script refuses to start logflare while vector.yml says otherwise"
     },
     {
       "id": "analytics-ready",