@luanpdd/kit-mcp 1.33.0 → 1.35.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (379) hide show
  1. package/LICENSE +21 -21
  2. package/README.md +168 -168
  3. package/gates/agent-no-recursive-dispatch.md +84 -84
  4. package/kit/COMANDOS.md +138 -138
  5. package/kit/COMPATIBILITY.md +70 -70
  6. package/kit/README.md +76 -76
  7. package/kit/agents/advisor-researcher.md +109 -109
  8. package/kit/agents/ai-mutation-tester.md +289 -289
  9. package/kit/agents/assumptions-analyzer.md +110 -110
  10. package/kit/agents/audit-log-implementer.md +314 -314
  11. package/kit/agents/auditor-consistencia-isolamento.md +414 -414
  12. package/kit/agents/b2b-saas-architect.md +157 -157
  13. package/kit/agents/burn-rate-forecaster.md +153 -153
  14. package/kit/agents/cascading-failures-auditor.md +299 -299
  15. package/kit/agents/codebase-mapper.md +769 -769
  16. package/kit/agents/crm-pipeline-implementer.md +257 -257
  17. package/kit/agents/debugger.md +814 -814
  18. package/kit/agents/designer-ui.md +216 -216
  19. package/kit/agents/detector-tenant-quente.md +338 -338
  20. package/kit/agents/evolution-go-integrator.md +201 -201
  21. package/kit/agents/example-reviewer.md +22 -22
  22. package/kit/agents/executor.md +565 -565
  23. package/kit/agents/golden-signals-instrumenter.md +232 -232
  24. package/kit/agents/incident-investigator.md +238 -238
  25. package/kit/agents/integration-checker.md +203 -203
  26. package/kit/agents/invite-flow-implementer.md +190 -190
  27. package/kit/agents/legacy-characterizer.md +369 -369
  28. package/kit/agents/lgpd-compliance-auditor.md +296 -296
  29. package/kit/agents/load-shedding-instrumenter.md +290 -290
  30. package/kit/agents/multi-tenant-isolation-auditor.md +254 -254
  31. package/kit/agents/multi-tenant-rls-writer.md +341 -341
  32. package/kit/agents/nyquist-auditor.md +181 -181
  33. package/kit/agents/observability-coverage-auditor.md +316 -316
  34. package/kit/agents/observability-instrumenter.md +191 -191
  35. package/kit/agents/omm-auditor.md +291 -291
  36. package/kit/agents/org-onboarding-implementer.md +224 -224
  37. package/kit/agents/payload-capture-instrumenter.md +274 -274
  38. package/kit/agents/phase-researcher.md +697 -697
  39. package/kit/agents/plan-checker.md +275 -275
  40. package/kit/agents/planner.md +923 -923
  41. package/kit/agents/postmortem-writer.md +273 -273
  42. package/kit/agents/project-researcher.md +653 -653
  43. package/kit/agents/prr-conductor.md +287 -287
  44. package/kit/agents/refactor-safety-auditor.md +405 -405
  45. package/kit/agents/release-pipeline-auditor.md +364 -364
  46. package/kit/agents/research-synthesizer.md +246 -246
  47. package/kit/agents/roadmapper.md +678 -678
  48. package/kit/agents/schema-checker.md +160 -160
  49. package/kit/agents/seam-finder.md +360 -360
  50. package/kit/agents/shotgun-surgery-detector.md +350 -350
  51. package/kit/agents/slo-engineer.md +217 -217
  52. package/kit/agents/storytelling-analyst.md +300 -300
  53. package/kit/agents/supabase-architect.md +249 -249
  54. package/kit/agents/supabase-auth-bootstrapper.md +400 -400
  55. package/kit/agents/supabase-auth-hook-writer.md +418 -418
  56. package/kit/agents/supabase-branching-architect.md +563 -563
  57. package/kit/agents/supabase-cicd-pipeline-implementer.md +778 -778
  58. package/kit/agents/supabase-column-privileges-writer.md +400 -400
  59. package/kit/agents/supabase-edge-fn-tester.md +288 -288
  60. package/kit/agents/supabase-edge-fn-writer.md +341 -341
  61. package/kit/agents/supabase-mfa-implementer.md +439 -439
  62. package/kit/agents/supabase-migration-writer.md +386 -386
  63. package/kit/agents/supabase-oauth-server-implementer.md +507 -507
  64. package/kit/agents/supabase-rbac-implementer.md +393 -393
  65. package/kit/agents/supabase-realtime-implementer.md +364 -364
  66. package/kit/agents/supabase-rls-hardener.md +522 -522
  67. package/kit/agents/supabase-rls-writer.md +324 -324
  68. package/kit/agents/supabase-roles-implementer.md +356 -356
  69. package/kit/agents/supabase-social-auth-implementer.md +451 -451
  70. package/kit/agents/supabase-sso-saml-architect.md +549 -549
  71. package/kit/agents/supabase-storage-implementer.md +407 -407
  72. package/kit/agents/super-admin-implementer.md +282 -282
  73. package/kit/agents/toil-auditor.md +268 -268
  74. package/kit/agents/ui-auditor.md +438 -438
  75. package/kit/agents/ui-checker.md +305 -305
  76. package/kit/agents/ui-researcher.md +356 -356
  77. package/kit/agents/user-profiler.md +176 -176
  78. package/kit/agents/validador-evolucao-schema.md +336 -336
  79. package/kit/agents/verifier.md +729 -729
  80. package/kit/agents/workflow-generator.md +167 -0
  81. package/kit/commands/adicionar-backlog.md +75 -75
  82. package/kit/commands/adicionar-fase.md +42 -42
  83. package/kit/commands/adicionar-tarefa.md +45 -45
  84. package/kit/commands/adicionar-testes.md +41 -41
  85. package/kit/commands/ajuda.md +21 -21
  86. package/kit/commands/atualizar.md +37 -37
  87. package/kit/commands/auditar-cascading.md +111 -111
  88. package/kit/commands/auditar-marco.md +179 -179
  89. package/kit/commands/auditar-observabilidade-cobertura-workflow.md +121 -0
  90. package/kit/commands/auditar-observabilidade-cobertura.md +183 -183
  91. package/kit/commands/auditar-refactor.md +219 -219
  92. package/kit/commands/auditar-release.md +109 -109
  93. package/kit/commands/auditar-uat.md +23 -23
  94. package/kit/commands/autonomo.md +40 -40
  95. package/kit/commands/branch-pr.md +24 -24
  96. package/kit/commands/burn-rate-status.md +408 -408
  97. package/kit/commands/capturar-payloads.md +193 -193
  98. package/kit/commands/caracterizar.md +212 -212
  99. package/kit/commands/concluir-marco.md +247 -247
  100. package/kit/commands/configuracoes.md +36 -36
  101. package/kit/commands/criar-workflow.md +158 -0
  102. package/kit/commands/dados-distribuidos.md +188 -188
  103. package/kit/commands/definir-perfil.md +10 -10
  104. package/kit/commands/depurar.md +190 -190
  105. package/kit/commands/detectar-duplicacao.md +197 -197
  106. package/kit/commands/discutir-fase.md +131 -131
  107. package/kit/commands/encontrar-seams.md +136 -136
  108. package/kit/commands/entrar-discord.md +17 -17
  109. package/kit/commands/estatisticas.md +18 -18
  110. package/kit/commands/example-greeting.md +33 -33
  111. package/kit/commands/executar-fase.md +58 -58
  112. package/kit/commands/expresso.md +56 -56
  113. package/kit/commands/fase-ui.md +34 -34
  114. package/kit/commands/fazer.md +57 -57
  115. package/kit/commands/fio.md +125 -125
  116. package/kit/commands/fluxos-trabalho.md +64 -64
  117. package/kit/commands/forense.md +176 -176
  118. package/kit/commands/gerenciador.md +38 -38
  119. package/kit/commands/inserir-fase.md +31 -31
  120. package/kit/commands/legacy.md +263 -263
  121. package/kit/commands/limpeza.md +17 -17
  122. package/kit/commands/listar-hipoteses-fase.md +45 -45
  123. package/kit/commands/listar-workspaces.md +18 -18
  124. package/kit/commands/load-shedding.md +117 -117
  125. package/kit/commands/mapear-codebase.md +70 -70
  126. package/kit/commands/multi-tenant.md +163 -163
  127. package/kit/commands/nota.md +33 -33
  128. package/kit/commands/novo-marco.md +43 -43
  129. package/kit/commands/novo-projeto.md +41 -41
  130. package/kit/commands/novo-workspace.md +43 -43
  131. package/kit/commands/pausar-trabalho.md +37 -37
  132. package/kit/commands/perfil-usuario.md +45 -45
  133. package/kit/commands/pesquisar-fase.md +195 -195
  134. package/kit/commands/planejar-fase.md +67 -67
  135. package/kit/commands/planejar-lacunas.md +33 -33
  136. package/kit/commands/plantar-ideia.md +25 -25
  137. package/kit/commands/progresso.md +24 -24
  138. package/kit/commands/proximo.md +30 -30
  139. package/kit/commands/publicar.md +490 -490
  140. package/kit/commands/rapido.md +35 -35
  141. package/kit/commands/reaplicar-patches.md +124 -124
  142. package/kit/commands/refactor-seguro.md +321 -321
  143. package/kit/commands/relatorio-sessao.md +19 -19
  144. package/kit/commands/remover-fase.md +31 -31
  145. package/kit/commands/remover-workspace.md +26 -26
  146. package/kit/commands/resumo-marco.md +50 -50
  147. package/kit/commands/retomar-trabalho.md +40 -40
  148. package/kit/commands/revisar-backlog.md +60 -60
  149. package/kit/commands/revisar-ui.md +32 -32
  150. package/kit/commands/revisar.md +37 -37
  151. package/kit/commands/saude.md +21 -21
  152. package/kit/commands/setup-notion.md +93 -93
  153. package/kit/commands/storytelling.md +179 -179
  154. package/kit/commands/supabase.md +238 -238
  155. package/kit/commands/sync-main.md +68 -68
  156. package/kit/commands/validar-fase.md +35 -35
  157. package/kit/commands/verificar-tarefas.md +44 -44
  158. package/kit/commands/verificar-trabalho.md +64 -64
  159. package/kit/file-manifest.json +424 -419
  160. package/kit/framework/bin/lib/commands.cjs +959 -959
  161. package/kit/framework/bin/lib/config.cjs +442 -442
  162. package/kit/framework/bin/lib/core.cjs +1230 -1230
  163. package/kit/framework/bin/lib/frontmatter.cjs +336 -336
  164. package/kit/framework/bin/lib/init.cjs +1442 -1442
  165. package/kit/framework/bin/lib/milestone.cjs +252 -252
  166. package/kit/framework/bin/lib/model-profiles.cjs +68 -68
  167. package/kit/framework/bin/lib/phase.cjs +888 -888
  168. package/kit/framework/bin/lib/profile-output.cjs +952 -952
  169. package/kit/framework/bin/lib/profile-pipeline.cjs +539 -539
  170. package/kit/framework/bin/lib/roadmap.cjs +329 -329
  171. package/kit/framework/bin/lib/security.cjs +382 -382
  172. package/kit/framework/bin/lib/state.cjs +1031 -1031
  173. package/kit/framework/bin/lib/template.cjs +222 -222
  174. package/kit/framework/bin/lib/uat.cjs +282 -282
  175. package/kit/framework/bin/lib/verify.cjs +888 -888
  176. package/kit/framework/bin/lib/workstream.cjs +491 -491
  177. package/kit/framework/bin/tools.cjs +918 -918
  178. package/kit/framework/commands/workstreams.md +63 -63
  179. package/kit/framework/references/checkpoints.md +778 -778
  180. package/kit/framework/references/continuation-format.md +249 -249
  181. package/kit/framework/references/decimal-phase-calculation.md +64 -64
  182. package/kit/framework/references/git-integration.md +295 -295
  183. package/kit/framework/references/git-planning-commit.md +38 -38
  184. package/kit/framework/references/model-profile-resolution.md +36 -36
  185. package/kit/framework/references/model-profiles.md +139 -139
  186. package/kit/framework/references/phase-argument-parsing.md +61 -61
  187. package/kit/framework/references/planning-config.md +202 -202
  188. package/kit/framework/references/questioning.md +162 -162
  189. package/kit/framework/references/tdd.md +263 -263
  190. package/kit/framework/references/ui-brand.md +160 -160
  191. package/kit/framework/references/user-profiling.md +657 -657
  192. package/kit/framework/references/verification-patterns.md +612 -612
  193. package/kit/framework/references/workstream-flag.md +58 -58
  194. package/kit/framework/templates/DEBUG.md +164 -164
  195. package/kit/framework/templates/UAT.md +265 -265
  196. package/kit/framework/templates/UI-SPEC.md +100 -100
  197. package/kit/framework/templates/VALIDATION.md +76 -76
  198. package/kit/framework/templates/claude-md.md +122 -122
  199. package/kit/framework/templates/codebase/architecture.md +185 -185
  200. package/kit/framework/templates/codebase/concerns.md +205 -205
  201. package/kit/framework/templates/codebase/conventions.md +204 -204
  202. package/kit/framework/templates/codebase/integrations.md +192 -192
  203. package/kit/framework/templates/codebase/stack.md +158 -158
  204. package/kit/framework/templates/codebase/structure.md +199 -199
  205. package/kit/framework/templates/codebase/testing.md +301 -301
  206. package/kit/framework/templates/config.json +44 -44
  207. package/kit/framework/templates/context.md +352 -352
  208. package/kit/framework/templates/continue-here.md +78 -78
  209. package/kit/framework/templates/copilot-instructions.md +7 -7
  210. package/kit/framework/templates/debug-subagent-prompt.md +91 -91
  211. package/kit/framework/templates/dev-preferences.md +20 -20
  212. package/kit/framework/templates/discovery.md +146 -146
  213. package/kit/framework/templates/discussion-log.md +63 -63
  214. package/kit/framework/templates/milestone-archive.md +123 -123
  215. package/kit/framework/templates/milestone.md +115 -115
  216. package/kit/framework/templates/phase-prompt.md +610 -610
  217. package/kit/framework/templates/planner-subagent-prompt.md +117 -117
  218. package/kit/framework/templates/project.md +186 -186
  219. package/kit/framework/templates/requirements.md +231 -231
  220. package/kit/framework/templates/research-project/ARCHITECTURE.md +204 -204
  221. package/kit/framework/templates/research-project/FEATURES.md +147 -147
  222. package/kit/framework/templates/research-project/PITFALLS.md +200 -200
  223. package/kit/framework/templates/research-project/STACK.md +120 -120
  224. package/kit/framework/templates/research-project/SUMMARY.md +170 -170
  225. package/kit/framework/templates/research.md +419 -419
  226. package/kit/framework/templates/retrospective.md +54 -54
  227. package/kit/framework/templates/roadmap.md +202 -202
  228. package/kit/framework/templates/state.md +176 -176
  229. package/kit/framework/templates/summary-complex.md +59 -59
  230. package/kit/framework/templates/summary-minimal.md +41 -41
  231. package/kit/framework/templates/summary-standard.md +48 -48
  232. package/kit/framework/templates/summary.md +209 -209
  233. package/kit/framework/templates/user-profile.md +146 -146
  234. package/kit/framework/templates/user-setup.md +256 -256
  235. package/kit/framework/templates/verification-report.md +258 -258
  236. package/kit/framework/workflows/add-phase.md +112 -112
  237. package/kit/framework/workflows/add-tests.md +351 -351
  238. package/kit/framework/workflows/add-todo.md +158 -158
  239. package/kit/framework/workflows/audit-milestone.md +340 -340
  240. package/kit/framework/workflows/audit-uat.md +109 -109
  241. package/kit/framework/workflows/autonomous.md +891 -891
  242. package/kit/framework/workflows/check-todos.md +177 -177
  243. package/kit/framework/workflows/cleanup.md +152 -152
  244. package/kit/framework/workflows/complete-milestone.md +696 -696
  245. package/kit/framework/workflows/diagnose-issues.md +231 -231
  246. package/kit/framework/workflows/discovery-phase.md +289 -289
  247. package/kit/framework/workflows/discuss-phase-assumptions.md +653 -653
  248. package/kit/framework/workflows/discuss-phase.md +784 -784
  249. package/kit/framework/workflows/do.md +104 -104
  250. package/kit/framework/workflows/execute-phase.md +838 -838
  251. package/kit/framework/workflows/execute-plan.md +510 -510
  252. package/kit/framework/workflows/fast.md +102 -102
  253. package/kit/framework/workflows/forensics.md +265 -265
  254. package/kit/framework/workflows/health.md +181 -181
  255. package/kit/framework/workflows/help.md +619 -619
  256. package/kit/framework/workflows/insert-phase.md +130 -130
  257. package/kit/framework/workflows/list-phase-assumptions.md +178 -178
  258. package/kit/framework/workflows/list-workspaces.md +56 -56
  259. package/kit/framework/workflows/manager.md +362 -362
  260. package/kit/framework/workflows/map-codebase.md +377 -377
  261. package/kit/framework/workflows/milestone-summary.md +223 -223
  262. package/kit/framework/workflows/new-milestone.md +486 -486
  263. package/kit/framework/workflows/new-project.md +1159 -1159
  264. package/kit/framework/workflows/new-workspace.md +237 -237
  265. package/kit/framework/workflows/next.md +97 -97
  266. package/kit/framework/workflows/node-repair.md +92 -92
  267. package/kit/framework/workflows/note.md +156 -156
  268. package/kit/framework/workflows/pause-work.md +176 -176
  269. package/kit/framework/workflows/plan-milestone-gaps.md +273 -273
  270. package/kit/framework/workflows/plan-phase.md +765 -765
  271. package/kit/framework/workflows/plant-seed.md +169 -169
  272. package/kit/framework/workflows/pr-branch.md +129 -129
  273. package/kit/framework/workflows/profile-user.md +450 -450
  274. package/kit/framework/workflows/progress.md +507 -507
  275. package/kit/framework/workflows/quick.md +757 -757
  276. package/kit/framework/workflows/remove-phase.md +155 -155
  277. package/kit/framework/workflows/remove-workspace.md +90 -90
  278. package/kit/framework/workflows/research-phase.md +82 -82
  279. package/kit/framework/workflows/resume-project.md +326 -326
  280. package/kit/framework/workflows/review.md +228 -228
  281. package/kit/framework/workflows/session-report.md +146 -146
  282. package/kit/framework/workflows/settings.md +283 -283
  283. package/kit/framework/workflows/ship.md +228 -228
  284. package/kit/framework/workflows/stats.md +60 -60
  285. package/kit/framework/workflows/transition.md +671 -671
  286. package/kit/framework/workflows/ui-phase.md +302 -302
  287. package/kit/framework/workflows/ui-review.md +165 -165
  288. package/kit/framework/workflows/update.md +323 -323
  289. package/kit/framework/workflows/validate-phase.md +174 -174
  290. package/kit/framework/workflows/verify-phase.md +252 -252
  291. package/kit/framework/workflows/verify-work.md +637 -637
  292. package/kit/hooks/check-update.js +118 -118
  293. package/kit/hooks/context-monitor.js +163 -163
  294. package/kit/hooks/kit-attribution-reminder.cjs +92 -92
  295. package/kit/hooks/kit-router.cjs +137 -137
  296. package/kit/hooks/prompt-guard.js +103 -103
  297. package/kit/hooks/statusline.js +125 -125
  298. package/kit/hooks/workflow-guard.js +101 -101
  299. package/kit/settings.json +45 -45
  300. package/kit/skills/ai-prompt-characterization/SKILL.md +335 -335
  301. package/kit/skills/armadilhas-sistemas-distribuidos/SKILL.md +447 -447
  302. package/kit/skills/audit-log-multi-tenant/SKILL.md +340 -340
  303. package/kit/skills/b2b-saas-architecture/SKILL.md +300 -300
  304. package/kit/skills/consistencia-leitura-replica/SKILL.md +385 -385
  305. package/kit/skills/crm-lead-pipeline-patterns/SKILL.md +343 -343
  306. package/kit/skills/dynamic-workflow-authoring/SKILL.md +223 -0
  307. package/kit/skills/escolha-modelo-consistencia/SKILL.md +494 -494
  308. package/kit/skills/evolucao-schema-compativel/SKILL.md +448 -448
  309. package/kit/skills/evolution-go-whatsapp-integration/SKILL.md +322 -322
  310. package/kit/skills/example-skill/SKILL.md +42 -42
  311. package/kit/skills/legacy-api-only-applications/SKILL.md +358 -358
  312. package/kit/skills/legacy-characterization-tests/SKILL.md +330 -330
  313. package/kit/skills/legacy-effect-analysis/SKILL.md +331 -331
  314. package/kit/skills/legacy-extract-class/SKILL.md +203 -203
  315. package/kit/skills/legacy-programming-by-difference/SKILL.md +252 -252
  316. package/kit/skills/legacy-seams-and-test-harness/SKILL.md +460 -460
  317. package/kit/skills/legacy-shotgun-surgery/SKILL.md +286 -286
  318. package/kit/skills/legacy-sprout-wrap-techniques/SKILL.md +434 -434
  319. package/kit/skills/legacy-storytelling-naked-crc/SKILL.md +270 -270
  320. package/kit/skills/lgpd-multi-tenant-compliance/SKILL.md +340 -340
  321. package/kit/skills/member-invite-flow/SKILL.md +305 -305
  322. package/kit/skills/member-management-react-shadcn/SKILL.md +328 -328
  323. package/kit/skills/multi-tenant-performance-scaling/SKILL.md +316 -316
  324. package/kit/skills/multi-tenant-rls-hierarchy/SKILL.md +342 -342
  325. package/kit/skills/org-onboarding-flow/SKILL.md +257 -257
  326. package/kit/skills/org-switcher-react-pattern/SKILL.md +349 -349
  327. package/kit/skills/permission-gate-react-pattern/SKILL.md +271 -271
  328. package/kit/skills/postgres-isolamento-concorrencia/SKILL.md +552 -552
  329. package/kit/skills/pre-refactor-characterization/SKILL.md +421 -421
  330. package/kit/skills/rbac-permissions-matrix-supabase/SKILL.md +338 -338
  331. package/kit/skills/streams-eventos-cdc/SKILL.md +711 -711
  332. package/kit/skills/supabase-auth-hardening/SKILL.md +674 -674
  333. package/kit/skills/supabase-auth-hooks/SKILL.md +875 -875
  334. package/kit/skills/supabase-auth-methods/SKILL.md +486 -486
  335. package/kit/skills/supabase-auth-sessions/SKILL.md +579 -579
  336. package/kit/skills/supabase-auth-ssr/SKILL.md +306 -306
  337. package/kit/skills/supabase-branching-workflow/SKILL.md +544 -544
  338. package/kit/skills/supabase-ci-cd-github-actions/SKILL.md +880 -880
  339. package/kit/skills/supabase-column-level-security/SKILL.md +426 -426
  340. package/kit/skills/supabase-config-toml-remotes/SKILL.md +807 -807
  341. package/kit/skills/supabase-custom-claims-rbac/SKILL.md +472 -472
  342. package/kit/skills/supabase-edge-functions/SKILL.md +330 -330
  343. package/kit/skills/supabase-edge-functions-auth/SKILL.md +309 -309
  344. package/kit/skills/supabase-edge-functions-limits/SKILL.md +302 -302
  345. package/kit/skills/supabase-edge-functions-mcp-server/SKILL.md +279 -279
  346. package/kit/skills/supabase-edge-functions-testing/SKILL.md +277 -277
  347. package/kit/skills/supabase-edge-runtime-builtins/SKILL.md +357 -357
  348. package/kit/skills/supabase-enterprise-sso-saml/SKILL.md +545 -545
  349. package/kit/skills/supabase-jwt-signing-keys/SKILL.md +399 -399
  350. package/kit/skills/supabase-mfa/SKILL.md +488 -488
  351. package/kit/skills/supabase-migration-repair/SKILL.md +823 -823
  352. package/kit/skills/supabase-migrations/SKILL.md +297 -297
  353. package/kit/skills/supabase-oauth-server/SKILL.md +537 -537
  354. package/kit/skills/supabase-pgtap-testing/SKILL.md +1053 -1053
  355. package/kit/skills/supabase-postgres-roles/SKILL.md +392 -392
  356. package/kit/skills/supabase-realtime/SKILL.md +460 -460
  357. package/kit/skills/supabase-rls-defense-in-depth/SKILL.md +418 -418
  358. package/kit/skills/supabase-rls-policies/SKILL.md +635 -635
  359. package/kit/skills/supabase-social-oauth/SKILL.md +480 -480
  360. package/kit/skills/supabase-third-party-auth/SKILL.md +450 -450
  361. package/kit/skills/super-admin-platform-pattern/SKILL.md +326 -326
  362. package/kit/skills/tenant-quente-mitigacao/SKILL.md +605 -605
  363. package/kit/skills/ui-anti-padroes-ia/SKILL.md +261 -261
  364. package/kit/skills/ui-contexto-produto/SKILL.md +248 -248
  365. package/kit/skills/ui-cor-estrategia/SKILL.md +213 -213
  366. package/kit/skills/ui-critica-auditoria/SKILL.md +260 -260
  367. package/kit/skills/ui-motion-funcional/SKILL.md +264 -264
  368. package/kit/skills/ui-ritmo-espacial/SKILL.md +259 -259
  369. package/kit/skills/ui-tipografia/SKILL.md +211 -211
  370. package/kit/skills/whatsapp-conversation-state-machine/SKILL.md +287 -287
  371. package/kit/workflows/auditar-observabilidade-cobertura.workflow.js +250 -0
  372. package/package.json +65 -63
  373. package/src/core/kit.js +333 -216
  374. package/src/core/reflect.js +247 -247
  375. package/src/core/registry.js +123 -112
  376. package/src/core/reverse-sync.js +448 -372
  377. package/src/core/sync.js +477 -437
  378. package/src/core/watch.js +121 -121
  379. package/src/mcp-server/index.js +794 -794
@@ -1,418 +1,418 @@
1
- ---
2
- name: supabase-rls-defense-in-depth
3
- description: Use ao desenhar defense-in-depth RLS em Supabase — event trigger rls_auto_enable (default projetos novos), BYPASSRLS role privilege, service_role caveat, security definer functions como bypass co…
4
- ---
5
-
6
- # Supabase — RLS Defense in Depth
7
-
8
- ## Quando usar
9
-
10
- LLM carrega esta skill quando precisar desenhar **camadas de defesa** RLS além das policies básicas. Trigger phrases:
11
-
12
- - "defense in depth RLS", "camadas de defesa Postgres"
13
- - "auto-enable RLS em todas tabelas novas", "event trigger ensure RLS"
14
- - "BYPASSRLS role privilege", "alter role bypassrls"
15
- - "security definer function bypassa RLS"
16
- - "service_role bypassa RLS", "service_role caveat"
17
- - "view security_invoker", "view bypass RLS"
18
- - "como proteger contra third-party tools acessando DB direto"
19
-
20
- ## Princípio canônico
21
-
22
- **Defense in depth** = múltiplas camadas independentes de proteção. Para RLS em Supabase:
23
-
24
- 1. **Camada 1 — Policy explícita por tabela** (skill `supabase-rls-policies`) — sempre granular, sempre `(select auth.uid())`, sempre `IS NOT NULL`, sempre indices.
25
- 2. **Camada 2 — Auto-enable RLS via event trigger** (DEFENSE-01) — garante que toda tabela nova nasce com RLS habilitado, mesmo se developer esquecer.
26
- 3. **Camada 3 — GRANT explícito** (skill `supabase-rls-policies`) — sem `grant select to authenticated`, queries falham com "permission denied" antes mesmo de chegar nas policies.
27
- 4. **Camada 4 — Bypass controlado** — `BYPASSRLS` role, `security definer` functions em schema `private` (DEFENSE-02, DEFENSE-04). Nunca em schema exposto.
28
- 5. **Camada 5 — Views com `security_invoker=true`** (DEFENSE-05, Postgres 15+) — views respeitam RLS do role chamador, não do criador.
29
- 6. **Camada 6 — Service role caveat** (DEFENSE-03) — entender que service_role bypassa RLS mas só no servidor; nunca expor ao cliente.
30
- 7. **Camada 7 — Cooperative handoff via supabase-rls-hardener** (v1.23) — todo SQL gerado pelo kit passa pelo hardener canonical antes do output final. Verdicts GO/STRENGTHEN/REWRITE-com-confirmação. Princípio canônico: agents externos pensam/planejam; agents Supabase materializam/hardenam.
31
- 8. **Camada 8 — Column-Level Privileges** (v1.24) — `GRANT/REVOKE (col1, col2) ON TABLE` para restringir colunas sensíveis (PII, audit payload, billing, tokens). Feature AVANÇADA — usar apenas quando RLS + dedicated role table não cobrem o caso. Cross-ref skill [`supabase-column-level-security`](../supabase-column-level-security/SKILL.md).
32
- 9. **Camada 9 — Auth Hooks - Custom Claims** (v1.25) — `Custom Access Token Auth Hook` (função PG `custom_access_token_hook(event jsonb)`) injeta `user_role` no JWT durante geração do token. RLS policies consultam o claim direto via `auth.jwt() ->> 'user_role'` ou via `authorize()` function — zero-JOIN, type-safe via enum, composable. Alternativa moderna a dedicated role table com JOIN custoso em policies. Caveat JWT freshness (eventually consistent). Cross-ref skill [`supabase-custom-claims-rbac`](../supabase-custom-claims-rbac/SKILL.md).
33
- 10. **Camada 10 — Postgres Roles Hierarchy** (v1.26) — Postgres roles dedicados para **system access** (service accounts, cron jobs, BI tools, ETL, admin scripts) em vez de service_role API key sempre. Auditabilidade superior (queries logam por role no `pg_stat_statements`), Role hierarchy via INHERIT/NOINHERIT, custom roles com BYPASSRLS específicos (`security_admin`, `dpo_role`, `lead_manager`, `platform_admin`). Distinção canônica vs application access (RLS + Custom Claims). Cross-ref skill [`supabase-postgres-roles`](../supabase-postgres-roles/SKILL.md).
34
-
35
- Razão: RLS é a primeira linha, mas humanos esquecem. Third-party tooling (Metabase, dbt, ferramentas BI conectadas via JDBC, scripts) bypassam toda a lógica da camada de aplicação — só RLS no banco protege. Defense in depth aplica princípio de **proteção sobreposta** para resiliência.
36
-
37
- ## DEFENSE-01: Event trigger `rls_auto_enable()` como default
38
-
39
- Em projetos novos, instale event trigger que ativa RLS automaticamente quando uma tabela é criada em schema exposto. Protege contra esquecimento humano.
40
-
41
- ### Função PLpgSQL + Event Trigger
42
-
43
- ```sql
44
- create or replace function rls_auto_enable()
45
- returns event_trigger
46
- language plpgsql
47
- security definer
48
- set search_path = pg_catalog
49
- as $$
50
- declare
51
- cmd record;
52
- begin
53
- for cmd in
54
- select *
55
- from pg_event_trigger_ddl_commands()
56
- where command_tag in ('CREATE TABLE', 'CREATE TABLE AS', 'SELECT INTO')
57
- and object_type in ('table','partitioned table')
58
- loop
59
- -- só atua em schemas configurados; pula sistema
60
- if cmd.schema_name is not null
61
- and cmd.schema_name in ('public')
62
- and cmd.schema_name not in ('pg_catalog','information_schema')
63
- and cmd.schema_name not like 'pg_toast%'
64
- and cmd.schema_name not like 'pg_temp%' then
65
- begin
66
- execute format('alter table if exists %s enable row level security', cmd.object_identity);
67
- raise log 'rls_auto_enable: enabled RLS on %', cmd.object_identity;
68
- exception
69
- when others then
70
- raise log 'rls_auto_enable: failed to enable RLS on %', cmd.object_identity;
71
- end;
72
- else
73
- raise log 'rls_auto_enable: skip % (system schema or not in enforced list)', cmd.object_identity;
74
- end if;
75
- end loop;
76
- end;
77
- $$;
78
-
79
- drop event trigger if exists ensure_rls;
80
- create event trigger ensure_rls
81
- on ddl_command_end
82
- when tag in ('CREATE TABLE', 'CREATE TABLE AS', 'SELECT INTO')
83
- execute function rls_auto_enable();
84
- ```
85
-
86
- ### Caveats e limitações
87
-
88
- - **Aplica-se a tabelas criadas após o trigger ser instalado** — tabelas pré-existentes precisam de `ALTER TABLE ... ENABLE ROW LEVEL SECURITY` manual ou script de migração.
89
- - **Schemas filtrados** — só atua em schemas listados explicitamente (`public` por default). Para incluir mais schemas, edite a lista. Para projetos multi-tenant em schemas separados por org, adicione `('public', 'org_*')` ou similar.
90
- - **Não cria policies** — apenas habilita RLS. Sem policies, tabela fica bloqueada para roles não-bypass — comportamento desejado (failsafe).
91
- - **Não dispara em `CREATE TABLE IF NOT EXISTS` quando tabela já existe** — comportamento padrão Postgres event triggers.
92
- - **Trigger é `SECURITY DEFINER`** — roda como o owner do trigger (geralmente `postgres`). Garante permissão para `ALTER TABLE`.
93
-
94
- ### Auditoria — listar tabelas sem RLS
95
-
96
- Para validar que defense-in-depth está aplicado:
97
-
98
- ```sql
99
- select schemaname, tablename
100
- from pg_tables
101
- where schemaname = 'public'
102
- and not exists (
103
- select 1 from pg_class c
104
- join pg_namespace n on n.oid = c.relnamespace
105
- where n.nspname = pg_tables.schemaname
106
- and c.relname = pg_tables.tablename
107
- and c.relrowsecurity = true
108
- );
109
- ```
110
-
111
- Se retorna ≥ 1 row, há tabelas sem RLS — gap de defense-in-depth.
112
-
113
- ## DEFENSE-02: `BYPASSRLS` role privilege para tarefas admin
114
-
115
- Para tarefas administrativas internas (jobs noturnos, migrations, scripts de manutenção), crie role com privilégio `BYPASSRLS` ao invés de usar `service_role`:
116
-
117
- ```sql
118
- -- criar role interno
119
- create role admin_internal with login password '<strong_pw>';
120
- alter role admin_internal with bypassrls;
121
-
122
- -- conceder permissões necessárias
123
- grant all on all tables in schema public to admin_internal;
124
- grant all on all sequences in schema public to admin_internal;
125
- ```
126
-
127
- **Quando usar:**
128
-
129
- - Scripts de manutenção que rodam dentro do banco (não via API)
130
- - Jobs `pg_cron` que precisam acessar dados cross-tenant
131
- - Migrations que populam dados em massa ignorando policies
132
- - DBAs investigando incidents em production
133
-
134
- **Quando NÃO usar:**
135
-
136
- - Edge Functions — use `SUPABASE_SERVICE_ROLE_KEY` direto (já tem BYPASSRLS implícito)
137
- - Backend customizado conectado via JDBC — use service_role role do Supabase
138
- - Qualquer role que receba requisição de cliente
139
-
140
- **Comparação com service_role:**
141
-
142
- | | `service_role` (Supabase API key) | Custom role `BYPASSRLS` |
143
- |---|---|---|
144
- | Escopo | API key — usado em Authorization header | Postgres role — login direto |
145
- | Auditoria | Logs do Supabase API | Postgres audit log (pg_audit) |
146
- | Revoke | Rotacionar key via dashboard | `revoke bypassrls from <role>` |
147
- | Granularidade | Tudo-ou-nada | Pode escopar GRANT a tabelas específicas |
148
-
149
- ## DEFENSE-03: `service_role` caveat — não bypassa RLS do user logged-in
150
-
151
- **Caveat crítico:** quando você usa o SDK Supabase com `SUPABASE_SERVICE_ROLE_KEY`, mas a sessão ainda tem `Authorization: Bearer <user_jwt>` set, **o RLS do user é aplicado** — o service_role bypass é **overridden**.
152
-
153
- ```js
154
- // errado — service_role mas com sessão user ainda ativa
155
- const supabase = createClient(URL, SERVICE_ROLE_KEY)
156
- supabase.auth.setSession({ access_token: userJwt, refresh_token: userRefresh })
157
- // agora .from('tasks').select() aplica RLS do user, não bypass
158
-
159
- // certo — service_role limpo
160
- const supabase = createClient(URL, SERVICE_ROLE_KEY, {
161
- auth: { persistSession: false, autoRefreshToken: false }
162
- })
163
- // agora .from('tasks').select() bypassa RLS
164
- ```
165
-
166
- **Como Supabase decide:** se a request tem JWT válido em Authorization header, Supabase deriva o `role` (anon/authenticated) e aplica RLS. Service_role só bypassa se NÃO há JWT user OU se você usa o pattern admin client com `persistSession: false`.
167
-
168
- **Aplicação:** Edge Functions que precisam fazer trabalho cross-tenant **devem** usar admin client separado (não o client da sessão do user que disparou a função):
169
-
170
- ```ts
171
- // edge function
172
- import { createClient } from 'npm:@supabase/supabase-js'
173
-
174
- Deno.serve(async (req) => {
175
- // admin client — bypassa RLS porque sem session
176
- const adminDb = createClient(
177
- Deno.env.get('SUPABASE_URL')!,
178
- Deno.env.get('SUPABASE_SERVICE_ROLE_KEY')!,
179
- { auth: { persistSession: false, autoRefreshToken: false } }
180
- )
181
-
182
- // user client — respeita RLS (para audit log do que user PODE fazer)
183
- const userDb = createClient(
184
- Deno.env.get('SUPABASE_URL')!,
185
- Deno.env.get('SUPABASE_ANON_KEY')!,
186
- { global: { headers: { Authorization: req.headers.get('Authorization')! } } }
187
- )
188
-
189
- // exemplo: validar com userDb (RLS aplicado) que user pode acessar org
190
- const { data: orgCheck } = await userDb.from('organizations').select('id').eq('id', orgId).single()
191
- if (!orgCheck) return new Response('Forbidden', { status: 403 })
192
-
193
- // executar mutation cross-tenant com adminDb (bypass RLS)
194
- const { data } = await adminDb.from('audit_log').insert({ ... })
195
- return new Response(JSON.stringify(data))
196
- })
197
- ```
198
-
199
- ## DEFENSE-04: `SECURITY DEFINER` functions como bypass controlado
200
-
201
- Funções `SECURITY DEFINER` rodam com permissões do owner (geralmente `postgres`, que tem `BYPASSRLS`). Use para encapsular lógica admin/cross-tenant que precisa bypassar RLS.
202
-
203
- ### Regras absolutas
204
-
205
- 1. **NUNCA em schema exposto** (`public`) — atacante poderia chamar a função com input arbitrário via REST API. Sempre em schema `private`, `internal`, ou similar **não-exposto** em API settings.
206
- 2. **Sempre `SET search_path = ''`** — evita schema injection (atacante criando função homônima em outro schema).
207
- 3. **Validar inputs** — função `SECURITY DEFINER` é alta autoridade; valide tudo (uuid format, range, ownership) antes de fazer trabalho.
208
- 4. **Auditar invocações** — log quem chamou + o quê + quando.
209
-
210
- ### Example: cross-tenant analytics aggregation
211
-
212
- ```sql
213
- -- schema private (NÃO exposto via API)
214
- create schema if not exists private;
215
-
216
- create or replace function private.org_analytics_summary(org_id_arg uuid)
217
- returns table(metric text, value bigint)
218
- language plpgsql
219
- security definer
220
- set search_path = ''
221
- as $$
222
- begin
223
- -- validar input (uuid format, ownership do caller, etc.)
224
- if org_id_arg is null then
225
- raise exception 'org_id_arg required';
226
- end if;
227
-
228
- -- validar que caller pode acessar essa org
229
- -- (mesmo bypassando RLS, regras de negócio aplicam)
230
- if not exists (
231
- select 1 from public.organization_members
232
- where org_id = org_id_arg and user_id = auth.uid()
233
- ) then
234
- raise exception 'access denied to org %', org_id_arg;
235
- end if;
236
-
237
- -- audit log
238
- insert into public.audit_log (event, user_id, org_id, payload)
239
- values ('analytics_summary_query', auth.uid(), org_id_arg, jsonb_build_object('ts', now()));
240
-
241
- -- query bypass RLS
242
- return query
243
- select 'total_tasks' as metric, count(*)::bigint as value
244
- from public.tasks
245
- where org_id = org_id_arg
246
- union all
247
- select 'total_members', count(*)::bigint
248
- from public.organization_members
249
- where org_id = org_id_arg;
250
- end;
251
- $$;
252
-
253
- -- expor via RPC (RLS check feito DENTRO da função, não via policy)
254
- grant execute on function private.org_analytics_summary(uuid) to authenticated;
255
- ```
256
-
257
- **Padrão de uso em policy:** funções `SECURITY DEFINER` podem ser chamadas em policies para fazer JOIN performant sem aplicar RLS recursivamente:
258
-
259
- ```sql
260
- -- ver supabase-rls-policies Performance section #5
261
- create function private.has_good_role()
262
- returns boolean
263
- language plpgsql
264
- security definer
265
- set search_path = ''
266
- as $$
267
- begin
268
- return exists (
269
- select 1 from public.roles_table
270
- where (select auth.uid()) = user_id and role = 'good_role'
271
- );
272
- end;
273
- $$;
274
-
275
- create policy "rls_test_select"
276
- on public.test_table for select
277
- to authenticated
278
- using ((select private.has_good_role()));
279
- ```
280
-
281
- ## DEFENSE-05: Views com `security_invoker=true` (Postgres 15+)
282
-
283
- Por padrão, views são criadas como `SECURITY DEFINER` — rodam com permissões do criador (geralmente `postgres`). Resultado: **views bypassam RLS** das tabelas subjacentes. Atacante pode contornar policies acessando a view ao invés da tabela.
284
-
285
- ### Postgres 15+: `security_invoker=true`
286
-
287
- ```sql
288
- -- view respeita RLS do role chamador (anon ou authenticated)
289
- create view public.user_active_tasks
290
- with (security_invoker = true)
291
- as
292
- select id, title, status, created_at
293
- from public.tasks
294
- where status = 'active';
295
- ```
296
-
297
- Agora `select * from user_active_tasks` aplica policies de `public.tasks` baseadas no `auth.uid()` do caller.
298
-
299
- ### Postgres < 15: revoke ou schema privado
300
-
301
- Em versões anteriores, `security_invoker` não está disponível. Mitigação:
302
-
303
- ```sql
304
- -- alternativa 1: revoke acesso de roles expostos
305
- revoke select on public.legacy_view from anon, authenticated;
306
- grant select on public.legacy_view to service_role; -- apenas backend
307
-
308
- -- alternativa 2: mover view para schema privado (não exposto via API)
309
- drop view if exists public.legacy_view;
310
- create view private.legacy_view as ...;
311
- -- não conceder a anon/authenticated; expor via RPC se necessário
312
- ```
313
-
314
- ### Auditoria — encontrar views vulneráveis
315
-
316
- ```sql
317
- -- views sem security_invoker em schemas expostos (Postgres 15+)
318
- select schemaname, viewname
319
- from pg_views v
320
- where schemaname = 'public'
321
- and not exists (
322
- select 1 from pg_class c
323
- join pg_namespace n on n.oid = c.relnamespace
324
- where n.nspname = v.schemaname
325
- and c.relname = v.viewname
326
- and c.reloptions::text like '%security_invoker=true%'
327
- );
328
- ```
329
-
330
- ## DEFENSE-06 (v1.24): Column-Level Privileges para PII/audit/billing/tokens
331
-
332
- Em tabelas com colunas sensíveis (PII em compliance LGPD/GDPR, audit log payload, billing data, tokens raw), aplique column-level privileges como **Camada 8** de defense-in-depth.
333
-
334
- ### Quando aplicar
335
-
336
- Use o checklist da skill [`supabase-column-level-security`](../supabase-column-level-security/SKILL.md):
337
-
338
- - **PII compliance:** SSN, CPF, salary, medical info
339
- - **Audit log sanitization:** `audit_log.payload` jsonb (legível só por security_admin role)
340
- - **Billing data:** `credit_card_token`, `bank_account`
341
- - **Tokens raw:** `org_invites.token_raw` (apenas service_role pós-create)
342
-
343
- ### Pattern canônico
344
-
345
- ```sql
346
- -- 1. REVOKE table-level (perde acesso a TODAS colunas)
347
- revoke select on table public.audit_log from authenticated;
348
-
349
- -- 2. GRANT column-level apenas em colunas não-sensíveis
350
- grant select (id, event_type, user_id, org_id, occurred_at)
351
- on table public.audit_log to authenticated;
352
-
353
- -- 3. service_role ou security_admin role mantém acesso total
354
- grant select on table public.audit_log to service_role;
355
- ```
356
-
357
- ### Caveat crítico — Wildcard `*` restriction
358
-
359
- Com column privileges, **`SELECT *` falha** — clientes devem listar colunas explicitamente:
360
-
361
- ```js
362
- // ❌ FALHA — wildcard expansion bate em colunas sem permission
363
- const { data } = supabase.from('audit_log').select()
364
-
365
- // ✅ OK — colunas explicitamente listadas
366
- const { data } = supabase.from('audit_log').select('id, event_type, user_id, org_id, occurred_at')
367
- ```
368
-
369
- ### Auditoria — detectar tabelas com PII sem column-level
370
-
371
- ```sql
372
- -- detectar colunas potencialmente sensíveis sem column-level GRANT/REVOKE
373
- select c.table_schema, c.table_name, c.column_name, c.data_type
374
- from information_schema.columns c
375
- where c.table_schema = 'public'
376
- and (
377
- c.column_name ilike any (array[
378
- '%email%', '%phone%', '%ssn%', '%cpf%', '%token%',
379
- '%password%', '%credit_card%', '%bank_account%', '%salary%'
380
- ])
381
- )
382
- and not exists (
383
- select 1 from information_schema.column_privileges p
384
- where p.table_schema = c.table_schema
385
- and p.table_name = c.table_name
386
- and p.column_name = c.column_name
387
- );
388
- ```
389
-
390
- Cross-ref auditoria sistemática em agent [`supabase-rls-hardener`](../../agents/supabase-rls-hardener.md) Detector 8 (v1.24) — invoca [`supabase-column-privileges-writer`](../../agents/supabase-column-privileges-writer.md) cooperativamente quando detecta gap.
391
-
392
- ## Resumo — checklist defense-in-depth (8 itens, v1.24)
393
-
394
- Use este checklist ao validar projetos Supabase em produção:
395
-
396
- - [ ] **DEFENSE-01:** Event trigger `rls_auto_enable` instalado em `ddl_command_end` cobrindo `CREATE TABLE` em schema `public`.
397
- - [ ] **DEFENSE-02:** Roles admin internos usam `BYPASSRLS` privilege ao invés de service_role API key em scripts/cron jobs.
398
- - [ ] **DEFENSE-03:** Edge Functions cross-tenant usam admin client separado (`persistSession: false`) ao invés de service_role com sessão user ativa.
399
- - [ ] **DEFENSE-04:** Lógica admin/cross-tenant encapsulada em funções `SECURITY DEFINER` em schema `private` (não-exposto), com `SET search_path = ''` + input validation + audit log.
400
- - [ ] **DEFENSE-05:** Views em Postgres 15+ usam `with (security_invoker = true)`; em versões anteriores, revoke acesso de `anon`/`authenticated` ou mover para schema privado.
401
- - [ ] **GRANT explícito** antes de ENABLE RLS em todas tabelas — sem `grant select to authenticated`, queries falham antes mesmo da policy.
402
- - [ ] **Cooperative handoff** — qualquer agent/skill/command produzindo SQL passa pelo `supabase-rls-hardener` (v1.23) antes do output final.
403
- - [ ] **DEFENSE-06 (v1.24):** Column-Level Privileges em tabelas com PII/audit payload/billing/tokens — REVOKE table-level + GRANT column-level granular; clientes listam colunas explicitamente (não `select *`).
404
- - [ ] **DEFENSE-07 (v1.25):** RBAC via Custom Access Token Auth Hook — `user_role` injetado no JWT durante geração do token; RLS policies consultam claim via `authorize()` function ao invés de JOIN em user_roles; `supabase_auth_admin` tem GRANT EXECUTE no hook + GRANT ALL em user_roles; revogação de role força logout via `auth.admin.signOut()` para invalidação imediata.
405
- - [ ] **DEFENSE-08 (v1.26):** Postgres Roles Hierarchy — service accounts internos (cron jobs, BI tools, ETL, admin scripts) usam Postgres roles dedicados em vez de service_role API key; custom roles com BYPASSRLS específicos (`security_admin`, `dpo_role`, `lead_manager`, `platform_admin`); role hierarchy via INHERIT/NOINHERIT para multi-level permissions; pg_stat_statements audit por role; cross-ref skill `supabase-postgres-roles` v1.26 para 10 predefined Supabase roles documentados.
406
-
407
- ## Cross-suite handoff cooperativo (v1.23)
408
-
409
- Esta skill é base para o agent `supabase-rls-hardener` (criado em Phase 126 do v1.23) — que recebe draft SQL via `Task()` upstream context, valida os 6 itens do checklist defense-in-depth acima, e devolve verdict GO/STRENGTHEN/REWRITE-com-confirmação.
410
-
411
- Princípio canônico v1.23: **agents não-Supabase pensam/planejam; agents Supabase materializam/hardenam; ninguém descarta upstream**. Esta skill consolida o conhecimento que o hardener aplica.
412
-
413
- ## Ver também
414
-
415
- - [supabase-rls-policies](../supabase-rls-policies/SKILL.md) — Camada 1 (policies granulares + IS NOT NULL + GRANT) + Performance recommendations
416
- - [supabase-migrations](../supabase-migrations/SKILL.md) — Template canônico v1.23 com 5 blocos obrigatórios (incluindo GRANT + ENABLE RLS + policies + index)
417
- - [supabase-database-functions](../supabase-database-functions/SKILL.md) — funções `SECURITY INVOKER` (default seguro) vs `SECURITY DEFINER` (com justificativa)
418
- - [glossário compartilhado](../_shared-supabase/glossary.md) — termos defense-in-depth, hardener, cooperative-handoff, event-trigger-rls-auto-enable, bypassrls, security_invoker
1
+ ---
2
+ name: supabase-rls-defense-in-depth
3
+ description: Use ao desenhar defense-in-depth RLS em Supabase — event trigger rls_auto_enable (default projetos novos), BYPASSRLS role privilege, service_role caveat, security definer functions como bypass co…
4
+ ---
5
+
6
+ # Supabase — RLS Defense in Depth
7
+
8
+ ## Quando usar
9
+
10
+ LLM carrega esta skill quando precisar desenhar **camadas de defesa** RLS além das policies básicas. Trigger phrases:
11
+
12
+ - "defense in depth RLS", "camadas de defesa Postgres"
13
+ - "auto-enable RLS em todas tabelas novas", "event trigger ensure RLS"
14
+ - "BYPASSRLS role privilege", "alter role bypassrls"
15
+ - "security definer function bypassa RLS"
16
+ - "service_role bypassa RLS", "service_role caveat"
17
+ - "view security_invoker", "view bypass RLS"
18
+ - "como proteger contra third-party tools acessando DB direto"
19
+
20
+ ## Princípio canônico
21
+
22
+ **Defense in depth** = múltiplas camadas independentes de proteção. Para RLS em Supabase:
23
+
24
+ 1. **Camada 1 — Policy explícita por tabela** (skill `supabase-rls-policies`) — sempre granular, sempre `(select auth.uid())`, sempre `IS NOT NULL`, sempre indices.
25
+ 2. **Camada 2 — Auto-enable RLS via event trigger** (DEFENSE-01) — garante que toda tabela nova nasce com RLS habilitado, mesmo se developer esquecer.
26
+ 3. **Camada 3 — GRANT explícito** (skill `supabase-rls-policies`) — sem `grant select to authenticated`, queries falham com "permission denied" antes mesmo de chegar nas policies.
27
+ 4. **Camada 4 — Bypass controlado** — `BYPASSRLS` role, `security definer` functions em schema `private` (DEFENSE-02, DEFENSE-04). Nunca em schema exposto.
28
+ 5. **Camada 5 — Views com `security_invoker=true`** (DEFENSE-05, Postgres 15+) — views respeitam RLS do role chamador, não do criador.
29
+ 6. **Camada 6 — Service role caveat** (DEFENSE-03) — entender que service_role bypassa RLS mas só no servidor; nunca expor ao cliente.
30
+ 7. **Camada 7 — Cooperative handoff via supabase-rls-hardener** (v1.23) — todo SQL gerado pelo kit passa pelo hardener canonical antes do output final. Verdicts GO/STRENGTHEN/REWRITE-com-confirmação. Princípio canônico: agents externos pensam/planejam; agents Supabase materializam/hardenam.
31
+ 8. **Camada 8 — Column-Level Privileges** (v1.24) — `GRANT/REVOKE (col1, col2) ON TABLE` para restringir colunas sensíveis (PII, audit payload, billing, tokens). Feature AVANÇADA — usar apenas quando RLS + dedicated role table não cobrem o caso. Cross-ref skill [`supabase-column-level-security`](../supabase-column-level-security/SKILL.md).
32
+ 9. **Camada 9 — Auth Hooks - Custom Claims** (v1.25) — `Custom Access Token Auth Hook` (função PG `custom_access_token_hook(event jsonb)`) injeta `user_role` no JWT durante geração do token. RLS policies consultam o claim direto via `auth.jwt() ->> 'user_role'` ou via `authorize()` function — zero-JOIN, type-safe via enum, composable. Alternativa moderna a dedicated role table com JOIN custoso em policies. Caveat JWT freshness (eventually consistent). Cross-ref skill [`supabase-custom-claims-rbac`](../supabase-custom-claims-rbac/SKILL.md).
33
+ 10. **Camada 10 — Postgres Roles Hierarchy** (v1.26) — Postgres roles dedicados para **system access** (service accounts, cron jobs, BI tools, ETL, admin scripts) em vez de service_role API key sempre. Auditabilidade superior (queries logam por role no `pg_stat_statements`), Role hierarchy via INHERIT/NOINHERIT, custom roles com BYPASSRLS específicos (`security_admin`, `dpo_role`, `lead_manager`, `platform_admin`). Distinção canônica vs application access (RLS + Custom Claims). Cross-ref skill [`supabase-postgres-roles`](../supabase-postgres-roles/SKILL.md).
34
+
35
+ Razão: RLS é a primeira linha, mas humanos esquecem. Third-party tooling (Metabase, dbt, ferramentas BI conectadas via JDBC, scripts) bypassam toda a lógica da camada de aplicação — só RLS no banco protege. Defense in depth aplica princípio de **proteção sobreposta** para resiliência.
36
+
37
+ ## DEFENSE-01: Event trigger `rls_auto_enable()` como default
38
+
39
+ Em projetos novos, instale event trigger que ativa RLS automaticamente quando uma tabela é criada em schema exposto. Protege contra esquecimento humano.
40
+
41
+ ### Função PLpgSQL + Event Trigger
42
+
43
+ ```sql
44
+ create or replace function rls_auto_enable()
45
+ returns event_trigger
46
+ language plpgsql
47
+ security definer
48
+ set search_path = pg_catalog
49
+ as $$
50
+ declare
51
+ cmd record;
52
+ begin
53
+ for cmd in
54
+ select *
55
+ from pg_event_trigger_ddl_commands()
56
+ where command_tag in ('CREATE TABLE', 'CREATE TABLE AS', 'SELECT INTO')
57
+ and object_type in ('table','partitioned table')
58
+ loop
59
+ -- só atua em schemas configurados; pula sistema
60
+ if cmd.schema_name is not null
61
+ and cmd.schema_name in ('public')
62
+ and cmd.schema_name not in ('pg_catalog','information_schema')
63
+ and cmd.schema_name not like 'pg_toast%'
64
+ and cmd.schema_name not like 'pg_temp%' then
65
+ begin
66
+ execute format('alter table if exists %s enable row level security', cmd.object_identity);
67
+ raise log 'rls_auto_enable: enabled RLS on %', cmd.object_identity;
68
+ exception
69
+ when others then
70
+ raise log 'rls_auto_enable: failed to enable RLS on %', cmd.object_identity;
71
+ end;
72
+ else
73
+ raise log 'rls_auto_enable: skip % (system schema or not in enforced list)', cmd.object_identity;
74
+ end if;
75
+ end loop;
76
+ end;
77
+ $$;
78
+
79
+ drop event trigger if exists ensure_rls;
80
+ create event trigger ensure_rls
81
+ on ddl_command_end
82
+ when tag in ('CREATE TABLE', 'CREATE TABLE AS', 'SELECT INTO')
83
+ execute function rls_auto_enable();
84
+ ```
85
+
86
+ ### Caveats e limitações
87
+
88
+ - **Aplica-se a tabelas criadas após o trigger ser instalado** — tabelas pré-existentes precisam de `ALTER TABLE ... ENABLE ROW LEVEL SECURITY` manual ou script de migração.
89
+ - **Schemas filtrados** — só atua em schemas listados explicitamente (`public` por default). Para incluir mais schemas, edite a lista. Para projetos multi-tenant em schemas separados por org, adicione `('public', 'org_*')` ou similar.
90
+ - **Não cria policies** — apenas habilita RLS. Sem policies, tabela fica bloqueada para roles não-bypass — comportamento desejado (failsafe).
91
+ - **Não dispara em `CREATE TABLE IF NOT EXISTS` quando tabela já existe** — comportamento padrão Postgres event triggers.
92
+ - **Trigger é `SECURITY DEFINER`** — roda como o owner do trigger (geralmente `postgres`). Garante permissão para `ALTER TABLE`.
93
+
94
+ ### Auditoria — listar tabelas sem RLS
95
+
96
+ Para validar que defense-in-depth está aplicado:
97
+
98
+ ```sql
99
+ select schemaname, tablename
100
+ from pg_tables
101
+ where schemaname = 'public'
102
+ and not exists (
103
+ select 1 from pg_class c
104
+ join pg_namespace n on n.oid = c.relnamespace
105
+ where n.nspname = pg_tables.schemaname
106
+ and c.relname = pg_tables.tablename
107
+ and c.relrowsecurity = true
108
+ );
109
+ ```
110
+
111
+ Se retorna ≥ 1 row, há tabelas sem RLS — gap de defense-in-depth.
112
+
113
+ ## DEFENSE-02: `BYPASSRLS` role privilege para tarefas admin
114
+
115
+ Para tarefas administrativas internas (jobs noturnos, migrations, scripts de manutenção), crie role com privilégio `BYPASSRLS` ao invés de usar `service_role`:
116
+
117
+ ```sql
118
+ -- criar role interno
119
+ create role admin_internal with login password '<strong_pw>';
120
+ alter role admin_internal with bypassrls;
121
+
122
+ -- conceder permissões necessárias
123
+ grant all on all tables in schema public to admin_internal;
124
+ grant all on all sequences in schema public to admin_internal;
125
+ ```
126
+
127
+ **Quando usar:**
128
+
129
+ - Scripts de manutenção que rodam dentro do banco (não via API)
130
+ - Jobs `pg_cron` que precisam acessar dados cross-tenant
131
+ - Migrations que populam dados em massa ignorando policies
132
+ - DBAs investigando incidents em production
133
+
134
+ **Quando NÃO usar:**
135
+
136
+ - Edge Functions — use `SUPABASE_SERVICE_ROLE_KEY` direto (já tem BYPASSRLS implícito)
137
+ - Backend customizado conectado via JDBC — use service_role role do Supabase
138
+ - Qualquer role que receba requisição de cliente
139
+
140
+ **Comparação com service_role:**
141
+
142
+ | | `service_role` (Supabase API key) | Custom role `BYPASSRLS` |
143
+ |---|---|---|
144
+ | Escopo | API key — usado em Authorization header | Postgres role — login direto |
145
+ | Auditoria | Logs do Supabase API | Postgres audit log (pg_audit) |
146
+ | Revoke | Rotacionar key via dashboard | `revoke bypassrls from <role>` |
147
+ | Granularidade | Tudo-ou-nada | Pode escopar GRANT a tabelas específicas |
148
+
149
+ ## DEFENSE-03: `service_role` caveat — não bypassa RLS do user logged-in
150
+
151
+ **Caveat crítico:** quando você usa o SDK Supabase com `SUPABASE_SERVICE_ROLE_KEY`, mas a sessão ainda tem `Authorization: Bearer <user_jwt>` set, **o RLS do user é aplicado** — o service_role bypass é **overridden**.
152
+
153
+ ```js
154
+ // errado — service_role mas com sessão user ainda ativa
155
+ const supabase = createClient(URL, SERVICE_ROLE_KEY)
156
+ supabase.auth.setSession({ access_token: userJwt, refresh_token: userRefresh })
157
+ // agora .from('tasks').select() aplica RLS do user, não bypass
158
+
159
+ // certo — service_role limpo
160
+ const supabase = createClient(URL, SERVICE_ROLE_KEY, {
161
+ auth: { persistSession: false, autoRefreshToken: false }
162
+ })
163
+ // agora .from('tasks').select() bypassa RLS
164
+ ```
165
+
166
+ **Como Supabase decide:** se a request tem JWT válido em Authorization header, Supabase deriva o `role` (anon/authenticated) e aplica RLS. Service_role só bypassa se NÃO há JWT user OU se você usa o pattern admin client com `persistSession: false`.
167
+
168
+ **Aplicação:** Edge Functions que precisam fazer trabalho cross-tenant **devem** usar admin client separado (não o client da sessão do user que disparou a função):
169
+
170
+ ```ts
171
+ // edge function
172
+ import { createClient } from 'npm:@supabase/supabase-js'
173
+
174
+ Deno.serve(async (req) => {
175
+ // admin client — bypassa RLS porque sem session
176
+ const adminDb = createClient(
177
+ Deno.env.get('SUPABASE_URL')!,
178
+ Deno.env.get('SUPABASE_SERVICE_ROLE_KEY')!,
179
+ { auth: { persistSession: false, autoRefreshToken: false } }
180
+ )
181
+
182
+ // user client — respeita RLS (para audit log do que user PODE fazer)
183
+ const userDb = createClient(
184
+ Deno.env.get('SUPABASE_URL')!,
185
+ Deno.env.get('SUPABASE_ANON_KEY')!,
186
+ { global: { headers: { Authorization: req.headers.get('Authorization')! } } }
187
+ )
188
+
189
+ // exemplo: validar com userDb (RLS aplicado) que user pode acessar org
190
+ const { data: orgCheck } = await userDb.from('organizations').select('id').eq('id', orgId).single()
191
+ if (!orgCheck) return new Response('Forbidden', { status: 403 })
192
+
193
+ // executar mutation cross-tenant com adminDb (bypass RLS)
194
+ const { data } = await adminDb.from('audit_log').insert({ ... })
195
+ return new Response(JSON.stringify(data))
196
+ })
197
+ ```
198
+
199
+ ## DEFENSE-04: `SECURITY DEFINER` functions como bypass controlado
200
+
201
+ Funções `SECURITY DEFINER` rodam com permissões do owner (geralmente `postgres`, que tem `BYPASSRLS`). Use para encapsular lógica admin/cross-tenant que precisa bypassar RLS.
202
+
203
+ ### Regras absolutas
204
+
205
+ 1. **NUNCA em schema exposto** (`public`) — atacante poderia chamar a função com input arbitrário via REST API. Sempre em schema `private`, `internal`, ou similar **não-exposto** em API settings.
206
+ 2. **Sempre `SET search_path = ''`** — evita schema injection (atacante criando função homônima em outro schema).
207
+ 3. **Validar inputs** — função `SECURITY DEFINER` é alta autoridade; valide tudo (uuid format, range, ownership) antes de fazer trabalho.
208
+ 4. **Auditar invocações** — log quem chamou + o quê + quando.
209
+
210
+ ### Example: cross-tenant analytics aggregation
211
+
212
+ ```sql
213
+ -- schema private (NÃO exposto via API)
214
+ create schema if not exists private;
215
+
216
+ create or replace function private.org_analytics_summary(org_id_arg uuid)
217
+ returns table(metric text, value bigint)
218
+ language plpgsql
219
+ security definer
220
+ set search_path = ''
221
+ as $$
222
+ begin
223
+ -- validar input (uuid format, ownership do caller, etc.)
224
+ if org_id_arg is null then
225
+ raise exception 'org_id_arg required';
226
+ end if;
227
+
228
+ -- validar que caller pode acessar essa org
229
+ -- (mesmo bypassando RLS, regras de negócio aplicam)
230
+ if not exists (
231
+ select 1 from public.organization_members
232
+ where org_id = org_id_arg and user_id = auth.uid()
233
+ ) then
234
+ raise exception 'access denied to org %', org_id_arg;
235
+ end if;
236
+
237
+ -- audit log
238
+ insert into public.audit_log (event, user_id, org_id, payload)
239
+ values ('analytics_summary_query', auth.uid(), org_id_arg, jsonb_build_object('ts', now()));
240
+
241
+ -- query bypass RLS
242
+ return query
243
+ select 'total_tasks' as metric, count(*)::bigint as value
244
+ from public.tasks
245
+ where org_id = org_id_arg
246
+ union all
247
+ select 'total_members', count(*)::bigint
248
+ from public.organization_members
249
+ where org_id = org_id_arg;
250
+ end;
251
+ $$;
252
+
253
+ -- expor via RPC (RLS check feito DENTRO da função, não via policy)
254
+ grant execute on function private.org_analytics_summary(uuid) to authenticated;
255
+ ```
256
+
257
+ **Padrão de uso em policy:** funções `SECURITY DEFINER` podem ser chamadas em policies para fazer JOIN performant sem aplicar RLS recursivamente:
258
+
259
+ ```sql
260
+ -- ver supabase-rls-policies Performance section #5
261
+ create function private.has_good_role()
262
+ returns boolean
263
+ language plpgsql
264
+ security definer
265
+ set search_path = ''
266
+ as $$
267
+ begin
268
+ return exists (
269
+ select 1 from public.roles_table
270
+ where (select auth.uid()) = user_id and role = 'good_role'
271
+ );
272
+ end;
273
+ $$;
274
+
275
+ create policy "rls_test_select"
276
+ on public.test_table for select
277
+ to authenticated
278
+ using ((select private.has_good_role()));
279
+ ```
280
+
281
+ ## DEFENSE-05: Views com `security_invoker=true` (Postgres 15+)
282
+
283
+ Por padrão, views são criadas como `SECURITY DEFINER` — rodam com permissões do criador (geralmente `postgres`). Resultado: **views bypassam RLS** das tabelas subjacentes. Atacante pode contornar policies acessando a view ao invés da tabela.
284
+
285
+ ### Postgres 15+: `security_invoker=true`
286
+
287
+ ```sql
288
+ -- view respeita RLS do role chamador (anon ou authenticated)
289
+ create view public.user_active_tasks
290
+ with (security_invoker = true)
291
+ as
292
+ select id, title, status, created_at
293
+ from public.tasks
294
+ where status = 'active';
295
+ ```
296
+
297
+ Agora `select * from user_active_tasks` aplica policies de `public.tasks` baseadas no `auth.uid()` do caller.
298
+
299
+ ### Postgres < 15: revoke ou schema privado
300
+
301
+ Em versões anteriores, `security_invoker` não está disponível. Mitigação:
302
+
303
+ ```sql
304
+ -- alternativa 1: revoke acesso de roles expostos
305
+ revoke select on public.legacy_view from anon, authenticated;
306
+ grant select on public.legacy_view to service_role; -- apenas backend
307
+
308
+ -- alternativa 2: mover view para schema privado (não exposto via API)
309
+ drop view if exists public.legacy_view;
310
+ create view private.legacy_view as ...;
311
+ -- não conceder a anon/authenticated; expor via RPC se necessário
312
+ ```
313
+
314
+ ### Auditoria — encontrar views vulneráveis
315
+
316
+ ```sql
317
+ -- views sem security_invoker em schemas expostos (Postgres 15+)
318
+ select schemaname, viewname
319
+ from pg_views v
320
+ where schemaname = 'public'
321
+ and not exists (
322
+ select 1 from pg_class c
323
+ join pg_namespace n on n.oid = c.relnamespace
324
+ where n.nspname = v.schemaname
325
+ and c.relname = v.viewname
326
+ and c.reloptions::text like '%security_invoker=true%'
327
+ );
328
+ ```
329
+
330
+ ## DEFENSE-06 (v1.24): Column-Level Privileges para PII/audit/billing/tokens
331
+
332
+ Em tabelas com colunas sensíveis (PII em compliance LGPD/GDPR, audit log payload, billing data, tokens raw), aplique column-level privileges como **Camada 8** de defense-in-depth.
333
+
334
+ ### Quando aplicar
335
+
336
+ Use o checklist da skill [`supabase-column-level-security`](../supabase-column-level-security/SKILL.md):
337
+
338
+ - **PII compliance:** SSN, CPF, salary, medical info
339
+ - **Audit log sanitization:** `audit_log.payload` jsonb (legível só por security_admin role)
340
+ - **Billing data:** `credit_card_token`, `bank_account`
341
+ - **Tokens raw:** `org_invites.token_raw` (apenas service_role pós-create)
342
+
343
+ ### Pattern canônico
344
+
345
+ ```sql
346
+ -- 1. REVOKE table-level (perde acesso a TODAS colunas)
347
+ revoke select on table public.audit_log from authenticated;
348
+
349
+ -- 2. GRANT column-level apenas em colunas não-sensíveis
350
+ grant select (id, event_type, user_id, org_id, occurred_at)
351
+ on table public.audit_log to authenticated;
352
+
353
+ -- 3. service_role ou security_admin role mantém acesso total
354
+ grant select on table public.audit_log to service_role;
355
+ ```
356
+
357
+ ### Caveat crítico — Wildcard `*` restriction
358
+
359
+ Com column privileges, **`SELECT *` falha** — clientes devem listar colunas explicitamente:
360
+
361
+ ```js
362
+ // ❌ FALHA — wildcard expansion bate em colunas sem permission
363
+ const { data } = supabase.from('audit_log').select()
364
+
365
+ // ✅ OK — colunas explicitamente listadas
366
+ const { data } = supabase.from('audit_log').select('id, event_type, user_id, org_id, occurred_at')
367
+ ```
368
+
369
+ ### Auditoria — detectar tabelas com PII sem column-level
370
+
371
+ ```sql
372
+ -- detectar colunas potencialmente sensíveis sem column-level GRANT/REVOKE
373
+ select c.table_schema, c.table_name, c.column_name, c.data_type
374
+ from information_schema.columns c
375
+ where c.table_schema = 'public'
376
+ and (
377
+ c.column_name ilike any (array[
378
+ '%email%', '%phone%', '%ssn%', '%cpf%', '%token%',
379
+ '%password%', '%credit_card%', '%bank_account%', '%salary%'
380
+ ])
381
+ )
382
+ and not exists (
383
+ select 1 from information_schema.column_privileges p
384
+ where p.table_schema = c.table_schema
385
+ and p.table_name = c.table_name
386
+ and p.column_name = c.column_name
387
+ );
388
+ ```
389
+
390
+ Cross-ref auditoria sistemática em agent [`supabase-rls-hardener`](../../agents/supabase-rls-hardener.md) Detector 8 (v1.24) — invoca [`supabase-column-privileges-writer`](../../agents/supabase-column-privileges-writer.md) cooperativamente quando detecta gap.
391
+
392
+ ## Resumo — checklist defense-in-depth (8 itens, v1.24)
393
+
394
+ Use este checklist ao validar projetos Supabase em produção:
395
+
396
+ - [ ] **DEFENSE-01:** Event trigger `rls_auto_enable` instalado em `ddl_command_end` cobrindo `CREATE TABLE` em schema `public`.
397
+ - [ ] **DEFENSE-02:** Roles admin internos usam `BYPASSRLS` privilege ao invés de service_role API key em scripts/cron jobs.
398
+ - [ ] **DEFENSE-03:** Edge Functions cross-tenant usam admin client separado (`persistSession: false`) ao invés de service_role com sessão user ativa.
399
+ - [ ] **DEFENSE-04:** Lógica admin/cross-tenant encapsulada em funções `SECURITY DEFINER` em schema `private` (não-exposto), com `SET search_path = ''` + input validation + audit log.
400
+ - [ ] **DEFENSE-05:** Views em Postgres 15+ usam `with (security_invoker = true)`; em versões anteriores, revoke acesso de `anon`/`authenticated` ou mover para schema privado.
401
+ - [ ] **GRANT explícito** antes de ENABLE RLS em todas tabelas — sem `grant select to authenticated`, queries falham antes mesmo da policy.
402
+ - [ ] **Cooperative handoff** — qualquer agent/skill/command produzindo SQL passa pelo `supabase-rls-hardener` (v1.23) antes do output final.
403
+ - [ ] **DEFENSE-06 (v1.24):** Column-Level Privileges em tabelas com PII/audit payload/billing/tokens — REVOKE table-level + GRANT column-level granular; clientes listam colunas explicitamente (não `select *`).
404
+ - [ ] **DEFENSE-07 (v1.25):** RBAC via Custom Access Token Auth Hook — `user_role` injetado no JWT durante geração do token; RLS policies consultam claim via `authorize()` function ao invés de JOIN em user_roles; `supabase_auth_admin` tem GRANT EXECUTE no hook + GRANT ALL em user_roles; revogação de role força logout via `auth.admin.signOut()` para invalidação imediata.
405
+ - [ ] **DEFENSE-08 (v1.26):** Postgres Roles Hierarchy — service accounts internos (cron jobs, BI tools, ETL, admin scripts) usam Postgres roles dedicados em vez de service_role API key; custom roles com BYPASSRLS específicos (`security_admin`, `dpo_role`, `lead_manager`, `platform_admin`); role hierarchy via INHERIT/NOINHERIT para multi-level permissions; pg_stat_statements audit por role; cross-ref skill `supabase-postgres-roles` v1.26 para 10 predefined Supabase roles documentados.
406
+
407
+ ## Cross-suite handoff cooperativo (v1.23)
408
+
409
+ Esta skill é base para o agent `supabase-rls-hardener` (criado em Phase 126 do v1.23) — que recebe draft SQL via `Task()` upstream context, valida os 6 itens do checklist defense-in-depth acima, e devolve verdict GO/STRENGTHEN/REWRITE-com-confirmação.
410
+
411
+ Princípio canônico v1.23: **agents não-Supabase pensam/planejam; agents Supabase materializam/hardenam; ninguém descarta upstream**. Esta skill consolida o conhecimento que o hardener aplica.
412
+
413
+ ## Ver também
414
+
415
+ - [supabase-rls-policies](../supabase-rls-policies/SKILL.md) — Camada 1 (policies granulares + IS NOT NULL + GRANT) + Performance recommendations
416
+ - [supabase-migrations](../supabase-migrations/SKILL.md) — Template canônico v1.23 com 5 blocos obrigatórios (incluindo GRANT + ENABLE RLS + policies + index)
417
+ - [supabase-database-functions](../supabase-database-functions/SKILL.md) — funções `SECURITY INVOKER` (default seguro) vs `SECURITY DEFINER` (com justificativa)
418
+ - [glossário compartilhado](../_shared-supabase/glossary.md) — termos defense-in-depth, hardener, cooperative-handoff, event-trigger-rls-auto-enable, bypassrls, security_invoker