@wildo-ai/framework-knowledge 1.1.0 → 1.1.2

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 (240) hide show
  1. package/dist/esm/framework-knowledge.definitions.d.ts +1 -1
  2. package/dist/esm/framework-knowledge.definitions.d.ts.map +1 -1
  3. package/dist/esm/framework-knowledge.definitions.js +43 -2
  4. package/dist/esm/framework-knowledge.definitions.js.map +1 -1
  5. package/dist/esm/framework-knowledge.loader.d.ts +19 -0
  6. package/dist/esm/framework-knowledge.loader.d.ts.map +1 -1
  7. package/dist/esm/framework-knowledge.loader.js +32 -0
  8. package/dist/esm/framework-knowledge.loader.js.map +1 -1
  9. package/dist/esm/framework-knowledge.schemas.d.ts +12 -0
  10. package/dist/esm/framework-knowledge.schemas.d.ts.map +1 -1
  11. package/dist/esm/framework-knowledge.schemas.js.map +1 -1
  12. package/dist/esm/playbooks/bottom_up.assemble-verify/playbook.definition.d.ts +46 -0
  13. package/dist/esm/playbooks/bottom_up.assemble-verify/playbook.definition.d.ts.map +1 -0
  14. package/dist/esm/playbooks/bottom_up.assemble-verify/playbook.definition.js +55 -0
  15. package/dist/esm/playbooks/bottom_up.assemble-verify/playbook.definition.js.map +1 -0
  16. package/dist/esm/playbooks/bottom_up.build-decompose/playbook.definition.d.ts +4 -4
  17. package/dist/esm/playbooks/bottom_up.build-decompose/playbook.definition.js +3 -3
  18. package/dist/esm/playbooks/bottom_up.build-decompose/playbook.definition.js.map +1 -1
  19. package/dist/esm/playbooks/bottom_up.build-dispatch/playbook.definition.d.ts +0 -3
  20. package/dist/esm/playbooks/bottom_up.build-dispatch/playbook.definition.d.ts.map +1 -1
  21. package/dist/esm/playbooks/bottom_up.build-dispatch/playbook.definition.js +0 -1
  22. package/dist/esm/playbooks/bottom_up.build-dispatch/playbook.definition.js.map +1 -1
  23. package/dist/esm/playbooks/bottom_up.build-scale/playbook.definition.d.ts +0 -3
  24. package/dist/esm/playbooks/bottom_up.build-scale/playbook.definition.d.ts.map +1 -1
  25. package/dist/esm/playbooks/bottom_up.build-scale/playbook.definition.js +0 -1
  26. package/dist/esm/playbooks/bottom_up.build-scale/playbook.definition.js.map +1 -1
  27. package/dist/esm/playbooks/bottom_up.domain-plan/playbook.definition.d.ts +2 -7
  28. package/dist/esm/playbooks/bottom_up.domain-plan/playbook.definition.d.ts.map +1 -1
  29. package/dist/esm/playbooks/bottom_up.domain-plan/playbook.definition.js +0 -1
  30. package/dist/esm/playbooks/bottom_up.domain-plan/playbook.definition.js.map +1 -1
  31. package/dist/esm/playbooks/bottom_up.foundation/playbook.definition.d.ts +8 -3
  32. package/dist/esm/playbooks/bottom_up.foundation/playbook.definition.d.ts.map +1 -1
  33. package/dist/esm/playbooks/bottom_up.foundation/playbook.definition.js +22 -1
  34. package/dist/esm/playbooks/bottom_up.foundation/playbook.definition.js.map +1 -1
  35. package/dist/esm/playbooks/bottom_up.placement-plan/playbook.definition.d.ts +0 -3
  36. package/dist/esm/playbooks/bottom_up.placement-plan/playbook.definition.d.ts.map +1 -1
  37. package/dist/esm/playbooks/bottom_up.placement-plan/playbook.definition.js +0 -1
  38. package/dist/esm/playbooks/bottom_up.placement-plan/playbook.definition.js.map +1 -1
  39. package/dist/esm/playbooks/refinement.compliance-document-candidate/playbook.definition.d.ts +4 -1
  40. package/dist/esm/playbooks/refinement.compliance-document-candidate/playbook.definition.d.ts.map +1 -1
  41. package/dist/esm/playbooks/refinement.compliance-document-candidate/playbook.definition.js +4 -1
  42. package/dist/esm/playbooks/refinement.compliance-document-candidate/playbook.definition.js.map +1 -1
  43. package/dist/esm/playbooks/refinement.document-facts/playbook.definition.d.ts +1 -1
  44. package/dist/esm/playbooks/refinement.document-facts/playbook.definition.js +13 -13
  45. package/dist/esm/playbooks/refinement.document-facts/playbook.definition.js.map +1 -1
  46. package/dist/esm/playbooks/top_down.brand-identity/playbook.definition.d.ts +3 -4
  47. package/dist/esm/playbooks/top_down.brand-identity/playbook.definition.d.ts.map +1 -1
  48. package/dist/esm/playbooks/top_down.brand-identity/playbook.definition.js +3 -3
  49. package/dist/esm/playbooks/top_down.brand-identity/playbook.definition.js.map +1 -1
  50. package/dist/esm/playbooks/top_down.business-model/playbook.definition.d.ts +2 -7
  51. package/dist/esm/playbooks/top_down.business-model/playbook.definition.d.ts.map +1 -1
  52. package/dist/esm/playbooks/top_down.business-model/playbook.definition.js +0 -1
  53. package/dist/esm/playbooks/top_down.business-model/playbook.definition.js.map +1 -1
  54. package/dist/esm/playbooks/top_down.capability-map/playbook.definition.d.ts +2 -7
  55. package/dist/esm/playbooks/top_down.capability-map/playbook.definition.d.ts.map +1 -1
  56. package/dist/esm/playbooks/top_down.capability-map/playbook.definition.js +0 -1
  57. package/dist/esm/playbooks/top_down.capability-map/playbook.definition.js.map +1 -1
  58. package/dist/esm/playbooks/top_down.compliance-analysis/playbook.definition.d.ts +54 -0
  59. package/dist/esm/playbooks/top_down.compliance-analysis/playbook.definition.d.ts.map +1 -0
  60. package/dist/esm/playbooks/top_down.compliance-analysis/playbook.definition.js +127 -0
  61. package/dist/esm/playbooks/top_down.compliance-analysis/playbook.definition.js.map +1 -0
  62. package/dist/esm/playbooks/top_down.compliance-conception/playbook.definition.d.ts +76 -0
  63. package/dist/esm/playbooks/top_down.compliance-conception/playbook.definition.d.ts.map +1 -0
  64. package/dist/esm/playbooks/top_down.compliance-conception/playbook.definition.js +92 -0
  65. package/dist/esm/playbooks/top_down.compliance-conception/playbook.definition.js.map +1 -0
  66. package/dist/esm/playbooks/top_down.compliance-evidence-adequacy/playbook.definition.d.ts +30 -0
  67. package/dist/esm/playbooks/top_down.compliance-evidence-adequacy/playbook.definition.d.ts.map +1 -0
  68. package/dist/esm/playbooks/top_down.compliance-evidence-adequacy/playbook.definition.js +29 -0
  69. package/dist/esm/playbooks/top_down.compliance-evidence-adequacy/playbook.definition.js.map +1 -0
  70. package/dist/esm/playbooks/top_down.compliance-posture/playbook.definition.d.ts +9 -44
  71. package/dist/esm/playbooks/top_down.compliance-posture/playbook.definition.d.ts.map +1 -1
  72. package/dist/esm/playbooks/top_down.compliance-posture/playbook.definition.js +4 -53
  73. package/dist/esm/playbooks/top_down.compliance-posture/playbook.definition.js.map +1 -1
  74. package/dist/esm/playbooks/top_down.compliance-research/playbook.definition.d.ts +31 -0
  75. package/dist/esm/playbooks/top_down.compliance-research/playbook.definition.d.ts.map +1 -0
  76. package/dist/esm/playbooks/top_down.compliance-research/playbook.definition.js +26 -0
  77. package/dist/esm/playbooks/top_down.compliance-research/playbook.definition.js.map +1 -0
  78. package/dist/esm/playbooks/top_down.market-analysis/playbook.definition.d.ts +2 -7
  79. package/dist/esm/playbooks/top_down.market-analysis/playbook.definition.d.ts.map +1 -1
  80. package/dist/esm/playbooks/top_down.market-analysis/playbook.definition.js +0 -1
  81. package/dist/esm/playbooks/top_down.market-analysis/playbook.definition.js.map +1 -1
  82. package/dist/esm/playbooks/top_down.market-snapshot-produce/playbook.definition.d.ts +0 -8
  83. package/dist/esm/playbooks/top_down.market-snapshot-produce/playbook.definition.d.ts.map +1 -1
  84. package/dist/esm/playbooks/top_down.market-snapshot-produce/playbook.definition.js +0 -1
  85. package/dist/esm/playbooks/top_down.market-snapshot-produce/playbook.definition.js.map +1 -1
  86. package/dist/esm/playbooks/top_down.phase-gate/playbook.definition.d.ts +2 -7
  87. package/dist/esm/playbooks/top_down.phase-gate/playbook.definition.d.ts.map +1 -1
  88. package/dist/esm/playbooks/top_down.phase-gate/playbook.definition.js +0 -1
  89. package/dist/esm/playbooks/top_down.phase-gate/playbook.definition.js.map +1 -1
  90. package/dist/esm/playbooks/top_down.requirements-definition/playbook.definition.d.ts +2 -7
  91. package/dist/esm/playbooks/top_down.requirements-definition/playbook.definition.d.ts.map +1 -1
  92. package/dist/esm/playbooks/top_down.requirements-definition/playbook.definition.js +0 -1
  93. package/dist/esm/playbooks/top_down.requirements-definition/playbook.definition.js.map +1 -1
  94. package/dist/esm/playbooks/top_down.scope-roadmap/playbook.definition.d.ts +2 -7
  95. package/dist/esm/playbooks/top_down.scope-roadmap/playbook.definition.d.ts.map +1 -1
  96. package/dist/esm/playbooks/top_down.scope-roadmap/playbook.definition.js +5 -3
  97. package/dist/esm/playbooks/top_down.scope-roadmap/playbook.definition.js.map +1 -1
  98. package/dist/esm/playbooks/top_down.value-proposition/playbook.definition.d.ts +0 -3
  99. package/dist/esm/playbooks/top_down.value-proposition/playbook.definition.d.ts.map +1 -1
  100. package/dist/esm/playbooks/top_down.value-proposition/playbook.definition.js +0 -1
  101. package/dist/esm/playbooks/top_down.value-proposition/playbook.definition.js.map +1 -1
  102. package/dist/esm/playbooks/top_down.vision-elicitation/playbook.definition.d.ts +4 -5
  103. package/dist/esm/playbooks/top_down.vision-elicitation/playbook.definition.d.ts.map +1 -1
  104. package/dist/esm/playbooks/top_down.vision-elicitation/playbook.definition.js +25 -4
  105. package/dist/esm/playbooks/top_down.vision-elicitation/playbook.definition.js.map +1 -1
  106. package/dist/knowledge/assurance-authority-bases/wildo-safe-baseline-v1/authority-basis.json +1 -1
  107. package/dist/knowledge/assurance-authority-bases/wildo-safe-baseline-v1/external-framework-mapping.json +1 -0
  108. package/dist/knowledge/manifest.json +8 -8
  109. package/dist/knowledge/playbooks/bottom_up.assemble-verify/method.md +51 -0
  110. package/dist/knowledge/playbooks/bottom_up.assemble-verify/playbook.json +43 -0
  111. package/dist/knowledge/playbooks/bottom_up.build-decompose/method.md +27 -10
  112. package/dist/knowledge/playbooks/bottom_up.build-decompose/playbook.json +8 -7
  113. package/dist/knowledge/playbooks/bottom_up.build-dispatch/method.md +6 -4
  114. package/dist/knowledge/playbooks/bottom_up.build-dispatch/playbook.json +4 -6
  115. package/dist/knowledge/playbooks/bottom_up.build-scale/method.md +5 -4
  116. package/dist/knowledge/playbooks/bottom_up.build-scale/playbook.json +4 -6
  117. package/dist/knowledge/playbooks/bottom_up.domain-plan/method.md +0 -4
  118. package/dist/knowledge/playbooks/bottom_up.domain-plan/playbook.json +4 -6
  119. package/dist/knowledge/playbooks/bottom_up.foundation/method.md +49 -22
  120. package/dist/knowledge/playbooks/bottom_up.foundation/playbook.json +14 -6
  121. package/dist/knowledge/playbooks/bottom_up.placement-plan/playbook.json +2 -4
  122. package/dist/knowledge/playbooks/refinement.compliance-document-candidate/method.md +46 -1
  123. package/dist/knowledge/playbooks/refinement.compliance-document-candidate/playbook.json +7 -3
  124. package/dist/knowledge/playbooks/refinement.document-facts/method.md +18 -4
  125. package/dist/knowledge/playbooks/refinement.document-facts/playbook.json +4 -2
  126. package/dist/knowledge/playbooks/top_down.brand-identity/method.md +2 -3
  127. package/dist/knowledge/playbooks/top_down.brand-identity/playbook.json +8 -7
  128. package/dist/knowledge/playbooks/top_down.business-model/playbook.json +3 -5
  129. package/dist/knowledge/playbooks/top_down.capability-map/method.md +0 -2
  130. package/dist/knowledge/playbooks/top_down.capability-map/playbook.json +4 -6
  131. package/dist/knowledge/playbooks/top_down.compliance-analysis/method.md +41 -0
  132. package/dist/knowledge/playbooks/top_down.compliance-analysis/playbook.json +100 -0
  133. package/dist/knowledge/playbooks/top_down.compliance-conception/method.md +117 -0
  134. package/dist/knowledge/playbooks/top_down.compliance-conception/playbook.json +86 -0
  135. package/dist/knowledge/playbooks/top_down.compliance-evidence-adequacy/method.md +61 -0
  136. package/dist/knowledge/playbooks/top_down.compliance-evidence-adequacy/playbook.json +41 -0
  137. package/dist/knowledge/playbooks/top_down.compliance-posture/method.md +21 -4
  138. package/dist/knowledge/playbooks/top_down.compliance-posture/playbook.json +14 -57
  139. package/dist/knowledge/playbooks/top_down.compliance-research/method.md +76 -0
  140. package/dist/knowledge/playbooks/top_down.compliance-research/playbook.json +42 -0
  141. package/dist/knowledge/playbooks/top_down.market-analysis/playbook.json +3 -5
  142. package/dist/knowledge/playbooks/top_down.market-evidence-adequacy/playbook.json +3 -1
  143. package/dist/knowledge/playbooks/top_down.market-research/playbook.json +3 -1
  144. package/dist/knowledge/playbooks/top_down.market-snapshot-produce/playbook.json +3 -5
  145. package/dist/knowledge/playbooks/top_down.phase-gate/method.md +8 -4
  146. package/dist/knowledge/playbooks/top_down.phase-gate/playbook.json +4 -6
  147. package/dist/knowledge/playbooks/top_down.requirements-definition/method.md +9 -7
  148. package/dist/knowledge/playbooks/top_down.requirements-definition/playbook.json +4 -6
  149. package/dist/knowledge/playbooks/top_down.scope-roadmap/method.md +2 -2
  150. package/dist/knowledge/playbooks/top_down.scope-roadmap/playbook.json +8 -7
  151. package/dist/knowledge/playbooks/top_down.value-proposition/method.md +0 -2
  152. package/dist/knowledge/playbooks/top_down.value-proposition/playbook.json +4 -6
  153. package/dist/knowledge/playbooks/top_down.vision-elicitation/method.md +2 -2
  154. package/dist/knowledge/playbooks/top_down.vision-elicitation/playbook.json +8 -7
  155. package/dist/knowledge/references/companion.technical-doc.service.md +102 -2
  156. package/dist/knowledge/references/example.config.app-shell.full.md +0 -1
  157. package/dist/knowledge/references/example.module.companion-exports.full.md +1 -0
  158. package/dist/knowledge/references/platform.config.saas.md +1 -0
  159. package/dist/knowledge/references/presets.models.surface-policy.md +19 -10
  160. package/dist/knowledge/references/saas.frontend.app-level-component.configuration.md +25 -1
  161. package/dist/knowledge/references/saas.frontend.composite-view.authoring.md +45 -0
  162. package/dist/knowledge/references/saas.frontend.display-layout.reading-measure.md +34 -0
  163. package/dist/knowledge/references/saas.frontend.navigation.operation-override.md +45 -0
  164. package/dist/knowledge/references/saas.frontend.public-shell.identity-mode.md +65 -0
  165. package/dist/knowledge/references/saas.frontend.resource-operation-frontend-config.md +89 -0
  166. package/dist/knowledge/references/saas.models.data-mode-read-list.md +31 -4
  167. package/dist/knowledge/references/saas.models.engine-capabilities.md +26 -6
  168. package/dist/knowledge/references/saas.models.file-upload-grant.md +179 -0
  169. package/dist/knowledge/references/saas.models.navigation.surface-vocabulary.md +124 -0
  170. package/dist/knowledge/references/saas.models.navigation.transition-vocabulary.md +99 -0
  171. package/dist/knowledge/references/saas.models.navigation.zone-declaration.md +127 -0
  172. package/dist/knowledge/references/saas.models.provider-capabilities.md +4 -3
  173. package/dist/knowledge/references/saas.models.resource-config.operation-variant-client-timeout.md +42 -0
  174. package/dist/knowledge/references/saas.specifications.application-authored-document-fact-specification.contract.md +29 -0
  175. package/dist/knowledge/references/saas.specifications.application-document-fact-specification.contract.md +11 -31
  176. package/dist/knowledge/references/saas.specifications.assurance-document-blueprint.contract.md +32 -2
  177. package/dist/knowledge/references/saas.specifications.assurance-program.contract.md +29 -6
  178. package/dist/knowledge/references/saas.specifications.companion-exports.contract.md +20 -0
  179. package/dist/knowledge/references/saas.specifications.resource-spec-config.md +22 -0
  180. package/dist/knowledge/references/zod.decorators.display-edit-modes.md +6 -0
  181. package/dist/knowledge/references/zod.decorators.field-spec-params.md +31 -5
  182. package/dist/knowledge/references/zod.decorators.per-primitive-ui.md +43 -0
  183. package/dist/knowledge/references/zod.decorators.primitive-ui.md +3 -0
  184. package/dist/knowledge/references/zod.decorators.shared-structural-params.md +15 -0
  185. package/dist/knowledge/registries/assurance-authority-bases.json +3 -2
  186. package/dist/knowledge/registries/framework-topics.json +99 -1
  187. package/dist/knowledge/registries/playbooks.json +81 -8
  188. package/dist/knowledge/registries/source-tags.json +757 -588
  189. package/dist/knowledge/skills/INDEX.md +3 -3
  190. package/dist/knowledge/skills/brand-logo-authoring/SKILL.md +2 -1
  191. package/dist/knowledge/skills/compliance-document-authoring/references/document-composition-governance.md +5 -1
  192. package/dist/knowledge/skills/compliance-document-authoring/references/framework-application-document-fact-specification-contract.md +66 -37
  193. package/dist/knowledge/skills/compliance-document-authoring/references/framework-assurance-document-blueprint-contract.md +40 -7
  194. package/dist/knowledge/skills/config-app-shell/SKILL.md +25 -1
  195. package/dist/knowledge/skills/config-app-shell/references/framework-app-shell-api.md +25 -1
  196. package/dist/knowledge/skills/config-app-shell/references/framework-public-shell-identity-api.md +124 -0
  197. package/dist/knowledge/skills/config-design-system/SKILL.md +23 -1
  198. package/dist/knowledge/skills/config-design-system/references/framework-surface-policy-api.md +19 -10
  199. package/dist/knowledge/skills/config-design-system/references/loading-placeholder-authoring.md +47 -0
  200. package/dist/knowledge/skills/config-navigation/SKILL.md +75 -2
  201. package/dist/knowledge/skills/config-navigation/references/framework-content-width-api.md +98 -0
  202. package/dist/knowledge/skills/config-navigation/references/framework-navigation-operation-override-api.md +54 -0
  203. package/dist/knowledge/skills/config-navigation/references/framework-navigation-vocabulary-api.md +232 -0
  204. package/dist/knowledge/skills/config-navigation/references/framework-navigation-zone-declaration-api.md +149 -0
  205. package/dist/knowledge/skills/config-navigation/references/how-navigation-resolves.md +114 -0
  206. package/dist/knowledge/skills/config-navigation/references/navigation-configuration-map.md +133 -0
  207. package/dist/knowledge/skills/config-saas/SKILL.md +19 -2
  208. package/dist/knowledge/skills/config-saas/references/compliance-declarations.md +111 -0
  209. package/dist/knowledge/skills/config-saas/references/framework-provider-capabilities-api.md +30 -9
  210. package/dist/knowledge/skills/config-saas/references/framework-wildo-saas-config-api.md +1 -0
  211. package/dist/knowledge/skills/flows-actors-a2a-exposure/SKILL.md +2 -1
  212. package/dist/knowledge/skills/flows-actors-a2a-exposure/references/framework-flows-actors-a2a-contract.md +4 -3
  213. package/dist/knowledge/skills/market-companion-prompt-enrichment/SKILL.md +20 -0
  214. package/dist/knowledge/skills/market-companion-prompt-enrichment/references/framework-companion-behavior-api.md +20 -0
  215. package/dist/knowledge/skills/module-companion-exports/SKILL.md +1 -0
  216. package/dist/knowledge/skills/module-companion-exports/references/companion-exports-by-package.md +20 -0
  217. package/dist/knowledge/skills/resource-add-field/SKILL.md +3 -3
  218. package/dist/knowledge/skills/resource-add-field/references/framework-field-cross-layer-api.md +104 -0
  219. package/dist/knowledge/skills/resource-config-wiring/SKILL.md +13 -0
  220. package/dist/knowledge/skills/resource-config-wiring/references/framework-saas-models-resource-config-api.md +71 -4
  221. package/dist/knowledge/skills/resource-custom-view/references/custom-view-contracts.md +89 -0
  222. package/dist/knowledge/skills/resource-mcp-tool-exposure/SKILL.md +22 -1
  223. package/dist/knowledge/skills/resource-mcp-tool-exposure/references/framework-file-upload-grant-contract.md +192 -0
  224. package/dist/knowledge/skills/resource-rag-source-fields/references/framework-embeddings-capability-api.md +28 -8
  225. package/dist/knowledge/skills/resource-schema-authoring/references/framework-shared-structural-params-api.md +15 -0
  226. package/dist/knowledge/skills/resource-ui-behavior/references/framework-display-edit-modes-api.md +6 -0
  227. package/dist/knowledge/skills/resource-ui-behavior/references/framework-frontend-resource-ui-contracts-api.md +89 -0
  228. package/dist/knowledge/skills/resource-ui-behavior/references/framework-per-primitive-ui-api.md +43 -0
  229. package/dist/knowledge/skills/resource-ui-behavior/references/framework-primitive-ui-api.md +3 -0
  230. package/dist/knowledge/skills/settings-placement-authoring/SKILL.md +1 -1
  231. package/dist/knowledge/skills/spec-companion-exports/SKILL.md +1 -0
  232. package/dist/knowledge/skills/spec-companion-exports/references/framework-companion-exports-contract-api.md +20 -0
  233. package/dist/knowledge/skills/spec-resource-update/references/framework-field-spec-params-api.md +31 -5
  234. package/dist/knowledge/skills/spec-resource-update/references/framework-spec-resource-contracts-api.md +22 -0
  235. package/dist/knowledge/skills/views-composite-authoring/references/framework-composite-view-contracts-api.md +45 -0
  236. package/dist/knowledge/topic-index.json +504 -156
  237. package/dist/tsconfig.build.tsbuildinfo +1 -1
  238. package/dist/tsconfig.tsbuildinfo +1 -1
  239. package/package.json +4 -4
  240. package/dist/esm/.builder.pid +0 -9
@@ -1,19 +1,23 @@
1
1
  {
2
2
  "ref": "top_down.compliance-posture",
3
- "version": "0.1.0",
3
+ "version": "0.2.1",
4
4
  "phase": "top_down",
5
- "intent": "Compliance-posture pass: a focused elicitation that determines which legal, regulatory, contractual and external-assurance regimes the product operates under by reasoning over its jurisdictions, the classes of data it handles, its sector and customer commitments — and records each applicable regime as a `compliance`-kind constraint (prose statement) on the product's VisionSnapshot (the `vision-brief` artifact family). This SHARPENS the compliance seam the vision-elicitation step named in passing; it does NOT re-elicit the whole vision. SCAFFOLD depth: compliance stays PROSE constraints here — the deep per-regime obligation model (control catalogs, data-residency matrices, obligation graphs) is a deferred later concern, not built here. v1 scaffold methodology — iterate the regime-derivation rubric.",
5
+ "intent": "Compliance-posture pass: a focused elicitation that determines which legal, regulatory, contractual and external-assurance regimes the product operates under by reasoning over its jurisdictions, the classes of data it handles, its sector and customer commitments — and records each applicable regime as a `compliance`-kind constraint (prose statement) on the product's VisionSnapshot (the `vision-brief` artifact family). This SHARPENS the compliance seam the vision-elicitation step named in passing; it does NOT re-elicit the whole vision. SCAFFOLD depth: compliance stays PROSE constraints here — the deep per-regime obligation model (control catalogs, data-residency matrices, obligation graphs) is a deferred later concern, not built here. v1 scaffold methodology — iterate the regime-derivation rubric. Its regimes used to be derived from nine interview answers plus a model's recollection; it now runs as the artifact-producing child of the `top_down.compliance-analysis` composite, only once an adequacy gate has read the gathered regime evidence and returned ADEQUATE.",
6
6
  "functionalTopics": [],
7
7
  "preconditions": [
8
- {
9
- "kind": "phase_is",
10
- "phase": "top_down"
11
- },
12
8
  {
13
9
  "kind": "artifact_exists",
14
10
  "selector": {
15
11
  "family": "vision-brief"
16
12
  }
13
+ },
14
+ {
15
+ "kind": "step_available",
16
+ "ref": "compliance-adequacy",
17
+ "requires": {
18
+ "path": "verdict",
19
+ "equals": "adequate"
20
+ }
17
21
  }
18
22
  ],
19
23
  "inputs": [
@@ -30,61 +34,14 @@
30
34
  ],
31
35
  "stepOutputs": [],
32
36
  "method": {
33
- "guidance": "# Compliance Posture — Derivation Method (v1 scaffold)\n\n### Target Application Framing\n\nThis knowledge is for configuring behavior in the target application.\nWildo is the framework surface being used, not the product being built.\n\nRead and apply these patterns from the application-developer point of view:\n\n- how the application should configure the relevant app-level file\n- how the application should use framework-provided factories, enums, and helpers\n- how the application should avoid mixing frontend, backend, and infrastructure concerns\n\n\n> **This is a v1 scaffold methodology.** The regime-derivation rubric below\n> (which triggers imply which regime, how much to capture) is a deliberately\n> explicit first cut meant to be tuned against real runs. Treat it as a\n> starting rubric, not a frozen contract. Crucially, it derives compliance\n> **posture at scaffold depth** — prose constraints, not a per-regime\n> obligation model.\n\n## What this playbook does\n\nYou determine the product's **compliance posture** — which regulatory regimes\nit must be built to satisfy — and record each as a `compliance`-kind constraint\non the existing vision. The vision-elicitation step already named the\ncompliance _seam_ in passing (a single \"we handle EU personal data\" line). Your\njob is the focused, deliberate pass over that seam: reason from the product's\njurisdictions, the classes of data it handles, and its sector to the concrete\nset of regimes, and write each one down as its own constraint with a clear\nprose statement.\n\nYou do **not** re-elicit the whole vision, design the domain, or draft\nrequirements. You touch exactly one slice of the `VisionSnapshot`: its\n`constraints[]` entries of `kind: compliance`. Everything else on the snapshot\n— the idea, the problem, the mission, the success criteria, the\nnon-compliance constraints — you leave exactly as you found it.\n\n## The scaffold-depth boundary (read this before you start)\n\nAt this stage compliance is captured as **prose constraints only**. Each\napplicable regime becomes one constraint entry whose `statement` names the\nregime and the data / jurisdiction it governs — e.g. _\"Must be GDPR-compliant\nfor EU/EEA customer personal data.\"_ That is the whole deliverable.\n\nDo **not** build, and do not try to author into the snapshot:\n\n- per-regime control catalogs or obligation graphs (legal-article mappings,\n sector safeguards, external-assurance criteria),\n- data-flow / residency matrices,\n- retention schedules, DPA registries, or audit-config derivations.\n\nThat deep per-regime obligation model is a deliberately deferred later concern.\nIf you catch yourself modelling controls or obligations, **stop** — you have\nleft scaffold depth. Naming the regime seam accurately here is exactly what\nlets that later, deeper step engage. Getting the _set of applicable regimes_\nright, in prose, is the win.\n\n### Transition to structured compliance conception\n\nThis playbook does not write the dedicated `compliance-program` family. Its\naccepted vision constraints are precursor facts for a later, separately\napproved compliance-conception step. That step may initialize or revise the\noptional structured family only after it resolves exact source versions and\nrights records, declarative application scope, applicability authority, Wildo\nobjective bindings, implementation choices, responsibilities, documentation\nrequirements and explicit gaps.\n\nThe transition is not an automatic conversion and must not reinterpret a\nvision sentence as a control mapping or positive compliance claim. Preserve\nthe vision constraint as product intent, record any missing detail as\n`UNDETERMINED` or an explicit gap, and accept the structured family through the\nnormal semantic-delta authority. Runtime organizations, assignees, evidence,\ncontrol occurrences and assessments remain outside both artifacts.\n\n## The derivation posture\n\n- **Derive, then confirm — do not just transcribe.** The typed questions on\n this playbook (`method.elicitation`) gather the raw triggers: jurisdictions,\n data classes, sector, and direct yes/no reads on the common regimes. Your\n value is reasoning from those triggers to the regime set, then reflecting\n each derived regime back to the creator for confirmation. If EU users +\n personal data are present, GDPR applies whether or not the creator named it;\n surface that, don't wait to be told.\n- **Regime derivation rubric (v1).** Use these as the starting triggers, and\n follow up where an answer is thin:\n - **GDPR / UK GDPR** — serving EU/EEA (or UK) individuals, or processing\n their personal data. Triggered by jurisdiction (EU/EEA/UK) + any personal\n data.\n - **HIPAA** — handling US protected health information as, or on behalf of, a\n covered entity or business associate. Triggered by health data + US\n healthcare context.\n - **External assurance commitments** — enterprise customers or the product's own security posture\n require an independent attestation or certification over security / availability /\n confidentiality. Often contractual rather than statutory — capture the requested target as\n externally supplied source metadata when the go-to-market depends on it.\n - **Data residency / sovereignty** — a requirement that certain data stay\n within a region or country. Triggered by residency answers or sector rules;\n record as its own compliance constraint even when it rides alongside GDPR.\n - **Sector & additional regimes** — payment-card, consumer-privacy, children's-data,\n government-security, contractual-assurance and sector-code obligations. Add each that\n the triggers imply; preserve the exact named source only in the later source registry.\n - When the triggers are genuinely absent (no personal data, no regulated\n sector), the honest posture is _no compliance constraints_ — say so and add\n none. Do not invent a regime to look thorough.\n- **One regime, one constraint.** Keep regimes as separate constraint entries\n rather than one omnibus \"must be compliant\" line — the set is read\n downstream, and a merged statement loses which regimes are actually in play.\n\n## What you emit\n\nYou refine the **existing** `VisionSnapshot` in place. Concretely, you\nadd or update entries in `constraints[]`:\n\n- Each applicable regime is one entry with `kind: compliance`.\n- Each entry has a stable kebab-case `ref` prefixed `constraint-`\n (e.g. `constraint-gdpr-eu-personal-data`,\n `constraint-hipaa-us-phi`, `constraint-external-assurance-security`,\n `constraint-data-residency-eu`), unique across the whole `constraints[]` set.\n- Each entry's `statement` is a prose sentence naming the regime and the data /\n jurisdiction it governs. Where the creator gave the source of the limit\n (a customer contract, a specific law), capture it as the constraint's\n rationale.\n\nLeave every non-`compliance` constraint, and every other vision entity,\nuntouched.\n\n**Idempotent — refine, do not duplicate.** Read the vision first. If compliance\nconstraints already exist (the vision-elicitation step, or a prior run of this\nplaybook, may have written some), reconcile against them: update a statement\nthat is now sharper, add a regime that was missed, and remove one only if the\ncreator confirms it never applied. **Match existing constraints by the REGIME\ntheir `statement` names, not by their `ref` slug** — the vision-elicitation step\nwrites compliance prose without following this playbook's ref convention, so a\nGDPR line it left may sit under any slug (`constraint-gdpr`,\n`constraint-eu-data`, …). When you refine a regime that is already present,\n**reuse that entry's existing `ref`** so it updates in place rather than\nforking; only mint a fresh `constraint-<regime>-<scope>` slug for a regime that\nis genuinely new. Re-running this playbook on an already-posture'd vision must\nconverge to one entry per regime, never accumulate near-duplicate\n`constraint-gdpr-*` entries.\n\n## Quality gate\n\nYour refined snapshot is checked by the deterministic `vision-snapshot-valid`\ngate (`validateVisionSnapshot`) before it commits — the same gate the\nvision-elicitation step uses, because you write to the same family. Duplicate\nrefs and malformed refs (a constraint ref that is not kebab-case with the\n`constraint-` prefix) are hard **errors**; the gate rejects them. Emit clean,\nunique refs the first time. Note that a vision with no compliance constraints\nis perfectly valid — an honestly unregulated product should pass the gate with\nnone.\n\n## Approval\n\nThe refined `VisionSnapshot` is presented to the creator for approval before it\nbecomes the committed `vision-brief`. Show the derived compliance posture back\nin plain prose — the list of regimes, each with the data / jurisdiction it\ngoverns and why it applies — and let the creator confirm or correct it.\nCompliance drives real downstream engine capability, audit, and retention\nchoices, so it is worth one explicit human sign-off that the regime set is\nright.\n\n## Runtime context\n\nBefore you start, read the current journey state and scope, and read the\nexisting vision so you refine its compliance constraints rather than writing a\nfresh, disconnected set.\n\n// TODO CLI COMMAND: wildo context journey-status (read current journey scope + artifacts)\n// TODO CLI COMMAND: wildo context vision-brief (read the existing vision + its current constraints)\n\n## References\n\n- `.claude/skills/journey-status/SKILL.md` — read current journey state and\n scope before deriving, and to see the `vision-brief` (and its existing\n compliance constraints) you refine.\n",
37
+ "guidance": "# Compliance Posture — Derivation Method (v1 scaffold)\n\n### Target Application Framing\n\nThis knowledge is for configuring behavior in the target application.\nWildo is the framework surface being used, not the product being built.\n\nRead and apply these patterns from the application-developer point of view:\n\n- how the application should configure the relevant app-level file\n- how the application should use framework-provided factories, enums, and helpers\n- how the application should avoid mixing frontend, backend, and infrastructure concerns\n\n\n> **This is a v1 scaffold methodology.** The regime-derivation rubric below\n> (which triggers imply which regime, how much to capture) is a deliberately\n> explicit first cut meant to be tuned against real runs. Treat it as a\n> starting rubric, not a frozen contract. Crucially, it derives compliance\n> **posture at scaffold depth** — prose constraints, not a per-regime\n> obligation model.\n\n## What this playbook does\n\nYou determine the product's **compliance posture** — which regulatory regimes\nit must be built to satisfy — and record each as a `compliance`-kind constraint\non the existing vision. The vision-elicitation step already named the\ncompliance _seam_ in passing (a single \"we handle EU personal data\" line). Your\njob is the focused, deliberate pass over that seam: reason from the product's\njurisdictions, the classes of data it handles, and its sector to the concrete\nset of regimes, and write each one down as its own constraint with a clear\nprose statement.\n\nYou do **not** re-elicit the whole vision, design the domain, or draft\nrequirements. You touch exactly one slice of the `VisionSnapshot`: its\n`constraints[]` entries of `kind: compliance`. Everything else on the snapshot\n— the idea, the problem, the mission, the success criteria, the\nnon-compliance constraints — you leave exactly as you found it.\n\n## The scaffold-depth boundary (read this before you start)\n\nAt this stage compliance is captured as **prose constraints only**. Each\napplicable regime becomes one constraint entry whose `statement` names the\nregime and the data / jurisdiction it governs — e.g. _\"Must be GDPR-compliant\nfor EU/EEA customer personal data.\"_ That is the whole deliverable.\n\nDo **not** build, and do not try to author into the snapshot:\n\n- per-regime control catalogs or obligation graphs (legal-article mappings,\n sector safeguards, external-assurance criteria),\n- data-flow / residency matrices,\n- retention schedules, DPA registries, or audit-config derivations.\n\nThat deep per-regime obligation model is a deliberately deferred later concern.\nIf you catch yourself modelling controls or obligations, **stop** — you have\nleft scaffold depth. Naming the regime seam accurately here is exactly what\nlets that later, deeper step engage. Getting the _set of applicable regimes_\nright, in prose, is the win.\n\n### Transition to structured compliance conception\n\nThis playbook does not write the dedicated `compliance-program` family. Its\naccepted vision constraints are precursor facts for a later, separately\napproved compliance-conception step. That step may initialize or revise the\noptional structured family only after it resolves exact source versions and\nrights records, declarative application scope, applicability authority, Wildo\nobjective bindings, implementation choices, responsibilities, documentation\nrequirements and explicit gaps.\n\nThe transition is not an automatic conversion and must not reinterpret a\nvision sentence as a control mapping or positive compliance claim. Preserve\nthe vision constraint as product intent, record any missing detail as\n`UNDETERMINED` or an explicit gap, and land the structured family as files in the\nspecifications package. Landing is not accepting: the run leaves working edits,\nand a human commit is the only acceptance there is. Runtime organizations, assignees, evidence,\ncontrol occurrences and assessments remain outside both artifacts.\n\n## The derivation posture\n\n- **Derive, then confirm — do not just transcribe.** The typed questions on\n this playbook (`method.elicitation`) gather the raw triggers: jurisdictions,\n data classes, sector, and direct yes/no reads on the common regimes. Your\n value is reasoning from those triggers to the regime set, then reflecting\n each derived regime back to the creator for confirmation. If EU users +\n personal data are present, GDPR applies whether or not the creator named it;\n surface that, don't wait to be told.\n- **Regime derivation rubric (v1).** Use these as the starting triggers, and\n follow up where an answer is thin:\n - **GDPR / UK GDPR** — serving EU/EEA (or UK) individuals, or processing\n their personal data. Triggered by jurisdiction (EU/EEA/UK) + any personal\n data.\n - **HIPAA** — handling US protected health information as, or on behalf of, a\n covered entity or business associate. Triggered by health data + US\n healthcare context.\n - **External assurance commitments** — enterprise customers or the product's own security posture\n require an independent attestation or certification over security / availability /\n confidentiality. Often contractual rather than statutory — capture the requested target as\n externally supplied source metadata when the go-to-market depends on it.\n - **Data residency / sovereignty** — a requirement that certain data stay\n within a region or country. Triggered by residency answers or sector rules;\n record as its own compliance constraint even when it rides alongside GDPR.\n - **Sector & additional regimes** — payment-card, consumer-privacy, children's-data,\n government-security, contractual-assurance and sector-code obligations. Add each that\n the triggers imply; preserve the exact named source only in the later source registry.\n - When the triggers are genuinely absent (no personal data, no regulated\n sector), the honest posture is _no compliance constraints_ — say so and add\n none. Do not invent a regime to look thorough.\n- **One regime, one constraint.** Keep regimes as separate constraint entries\n rather than one omnibus \"must be compliant\" line — the set is read\n downstream, and a merged statement loses which regimes are actually in play.\n\n## What you emit\n\nYou refine the **existing** `VisionSnapshot` in place. Concretely, you\nadd or update entries in `constraints[]`:\n\n- Each applicable regime is one entry with `kind: compliance`.\n- Each entry has a stable kebab-case `ref` prefixed `constraint-`\n (e.g. `constraint-gdpr-eu-personal-data`,\n `constraint-hipaa-us-phi`, `constraint-external-assurance-security`,\n `constraint-data-residency-eu`), unique across the whole `constraints[]` set.\n- Each entry's `statement` is a prose sentence naming the regime and the data /\n jurisdiction it governs. Where the creator gave the source of the limit\n (a customer contract, a specific law), capture it as the constraint's\n rationale.\n\nLeave every non-`compliance` constraint, and every other vision entity,\nuntouched.\n\n**Idempotent — refine, do not duplicate.** Read the vision first. If compliance\nconstraints already exist (the vision-elicitation step, or a prior run of this\nplaybook, may have written some), reconcile against them: update a statement\nthat is now sharper, add a regime that was missed, and remove one only if the\ncreator confirms it never applied. **Match existing constraints by the REGIME\ntheir `statement` names, not by their `ref` slug** — the vision-elicitation step\nwrites compliance prose without following this playbook's ref convention, so a\nGDPR line it left may sit under any slug (`constraint-gdpr`,\n`constraint-eu-data`, …). When you refine a regime that is already present,\n**reuse that entry's existing `ref`** so it updates in place rather than\nforking; only mint a fresh `constraint-<regime>-<scope>` slug for a regime that\nis genuinely new. Re-running this playbook on an already-posture'd vision must\nconverge to one entry per regime, never accumulate near-duplicate\n`constraint-gdpr-*` entries.\n\n## Quality gate\n\nYour refined snapshot is checked by the deterministic `vision-snapshot-valid`\ngate (`validateVisionSnapshot`) before it commits — the same gate the\nvision-elicitation step uses, because you write to the same family. Duplicate\nrefs and malformed refs (a constraint ref that is not kebab-case with the\n`constraint-` prefix) are hard **errors**; the gate rejects them. Emit clean,\nunique refs the first time. Note that a vision with no compliance constraints\nis perfectly valid — an honestly unregulated product should pass the gate with\nnone.\n\n## Approval\n\nThe refined `VisionSnapshot` is presented to the creator for approval before it\nbecomes the committed `vision-brief`. Show the derived compliance posture back\nin plain prose — the list of regimes, each with the data / jurisdiction it\ngoverns and why it applies — and let the creator confirm or correct it.\nCompliance drives real downstream engine capability, audit, and retention\nchoices, so it is worth one explicit human sign-off that the regime set is\nright.\n\n## Runtime context\n\nBefore you start, read the current journey state and scope, and read the\nexisting vision so you refine its compliance constraints rather than writing a\nfresh, disconnected set.\n\n(The existing vision and its current constraints are already in your brief as the `vision-brief`\ninput section.)\n\n## The evidence, and how to use it\n\nYour run carries what sibling steps gathered — source-attributed regime evidence and the adequacy\nverdict that admitted this step. Derive each constraint from the creator's answers AND that\nevidence, never from recollection:\n\n- A regime the evidence establishes as APPLICABLE becomes a `compliance` constraint whose statement\n names the regime, the trigger (jurisdiction / data class / sector / commitment), and what it\n governs — in the evidence's terms.\n- A regime the evidence EXCLUDES is not written as a constraint, and is not silently dropped either:\n where the creator's answers suggested it, the statement of a neighbouring constraint may note the\n exclusion so the decision is legible.\n- A trigger the evidence could NOT establish becomes a constraint stating the OPEN QUESTION, never a\n guessed answer — an unestablished regime stated as settled is the defect this composite exists to\n end.\n\n## References\n\n- `.claude/skills/journey-status/SKILL.md` — read current journey state and\n scope before deriving, and to see the `vision-brief` (and its existing\n compliance constraints) you refine.\n",
34
38
  "skillRefs": [
35
39
  "journey-status"
36
40
  ],
37
- "elicitation": {
38
- "questions": [
39
- {
40
- "ref": "handles-regulated-data",
41
- "question": "Does the product store, process, or transmit personal data or otherwise regulated data (health, financial, biometric, children's, government) — either from end users or on behalf of customers?",
42
- "kind": "confirm"
43
- },
44
- {
45
- "ref": "data-classes",
46
- "question": "Which classes of sensitive data does the product touch? List them (e.g. names/emails, health records, payment card data, government IDs, precise location, employee data). Answer \"none\" if it handles no personal or sensitive data.",
47
- "kind": "free_text"
48
- },
49
- {
50
- "ref": "jurisdictions",
51
- "question": "In which jurisdictions do your users live and your data reside or transit (e.g. EU/EEA, UK, US, California, Canada, specific countries)? This is what drives which privacy regimes apply.",
52
- "kind": "free_text"
53
- },
54
- {
55
- "ref": "sector",
56
- "question": "Does the product operate in a regulated sector with its own rules (healthcare, finance/banking, insurance, education, government, telecoms)? Name the sector, or answer \"none\".",
57
- "kind": "free_text"
58
- },
59
- {
60
- "ref": "gdpr-applies",
61
- "question": "Do you serve EU/EEA (or UK) individuals or process their personal data? (This is the primary trigger for GDPR / UK GDPR.)",
62
- "kind": "confirm"
63
- },
64
- {
65
- "ref": "hipaa-applies",
66
- "question": "Do you handle US protected health information as, or on behalf of, a covered entity or business associate? (This is the primary trigger for HIPAA.)",
67
- "kind": "confirm"
68
- },
69
- {
70
- "ref": "external-assurance-applies",
71
- "question": "Do enterprise customers (or your own security posture) require an independent attestation or certification covering security / availability / confidentiality of the service? Name the requested target only as an external source requirement.",
72
- "kind": "confirm"
73
- },
74
- {
75
- "ref": "data-residency",
76
- "question": "Are there data-residency or sovereignty requirements — must certain data stay within a region or country (e.g. EU-only hosting, in-country storage)? Describe them, or answer \"none\".",
77
- "kind": "free_text"
78
- },
79
- {
80
- "ref": "additional-regimes",
81
- "question": "Any other legal regimes, sector frameworks, government programs, or contractual security commitments that apply? List the externally named sources, or answer \"none\".",
82
- "kind": "free_text"
83
- }
84
- ]
85
- },
86
41
  "contextSources": [],
87
- "suspendsForClarification": false
42
+ "suspendsForClarification": false,
43
+ "briefAdequacyGate": false,
44
+ "briefAdequacyExcludedSources": []
88
45
  },
89
46
  "executor": "companion_agent",
90
47
  "approval": "approve_result",
@@ -0,0 +1,76 @@
1
+ # Compliance Research — gather the regime evidence
2
+
3
+ ### Target Application Framing
4
+
5
+ This knowledge is for configuring behavior in the target application.
6
+ Wildo is the framework surface being used, not the product being built.
7
+
8
+ Read and apply these patterns from the application-developer point of view:
9
+
10
+ - how the application should configure the relevant app-level file
11
+ - how the application should use framework-provided factories, enums, and helpers
12
+ - how the application should avoid mixing frontend, backend, and infrastructure concerns
13
+
14
+
15
+ You are the **researcher**. Your one job is to find out which regimes plausibly
16
+ apply to this product and report it with sources. You do not decide the posture,
17
+ author constraints, or touch any file — a sibling does that, from what you bring
18
+ back.
19
+
20
+ ## Why this step exists
21
+
22
+ The playbook that records the compliance posture used to derive it from nine
23
+ interview answers plus what the model remembered about the law. Recollection is
24
+ confident, fluent, and wrong in exactly the cases that matter here: thresholds
25
+ change, regimes gain territorial reach, sector rules turn on definitions no one
26
+ remembers precisely — and a wrongly excluded regime is a legal statement about the
27
+ product, not a stale price.
28
+
29
+ This step exists so the posture rests on pages someone actually read.
30
+
31
+ ## What counts as a regime
32
+
33
+ A regime is **public law or a regulator's instrument** — a statute, a regulation, a
34
+ directive as transposed, a supervisory authority's binding guidance. Those are what
35
+ you research, name and quote: they bind the product whether or not anyone agreed to
36
+ them, so applicability is a question of fact you can settle from a source.
37
+
38
+ A **voluntary security certification or attestation framework is not a regime.** It
39
+ binds only where the creator has committed to it, which makes it an elicited fact
40
+ rather than a researched one — so there is nothing for you to establish and no
41
+ threshold to test. Report the whole class as one line ("voluntary
42
+ security-certification and attestation frameworks — commitments, not legal
43
+ obligations; none committed" or whatever the creator's answers say), and do not name
44
+ an individual proprietary standard or reproduce any of its control identifiers: those
45
+ catalogues are licensed material, and this evidence store is committed into the
46
+ application's repository.
47
+
48
+ The distinction is not bookkeeping. A list that mixes the two invites a posture to
49
+ read a certification as an obligation the product is failing, which is a false legal
50
+ statement about the product — the exact defect this lane exists to prevent, arriving
51
+ from the other direction.
52
+
53
+ ## What to establish
54
+
55
+ Work the objective you were given — it was composed for THIS round, and on a later
56
+ round it names what the earlier rounds could not establish. Where it is broad,
57
+ these are the questions worth answering, per candidate regime:
58
+
59
+ 1. **The applicability trigger, exactly.** What makes it apply — territorial scope,
60
+ data classes, sector, size thresholds, customer commitments — quoted or closely
61
+ paraphrased from an authoritative source, with enough precision that a reader
62
+ can test THIS product against it.
63
+ 2. **Whether THIS product trips it**, given the vision's jurisdictions, data
64
+ classes and sector — stated as your reading, distinct from the trigger itself.
65
+ 3. **What the regime obliges at posture level.** One or two sentences — the deep
66
+ obligation model is a later concern; the posture needs "applies / does not
67
+ apply / cannot be established, because …".
68
+
69
+ ## How to report
70
+
71
+ - **Authoritative sources first.** A regulator's page or the official text over any
72
+ vendor summary; name the source beside every claim, with its URL.
73
+ - **Say what you could NOT establish.** The gaps aim the next round; hiding them
74
+ produces a posture that reads complete and is not — the exact defect this lane
75
+ replaces.
76
+ - **Never pad.** A regime you found no trigger for is one line, not a paragraph.
@@ -0,0 +1,42 @@
1
+ {
2
+ "ref": "top_down.compliance-research",
3
+ "version": "0.1.0",
4
+ "phase": "top_down",
5
+ "intent": "Gather EVIDENCE about which legal and regulatory regimes plausibly apply to this product — the applicability trigger of each candidate regime (jurisdiction, data class, sector), verified against authoritative sources — and return a condensed, source-attributed brief. A voluntary security certification or attestation is NOT a regime and is not researched here: it binds only if the creator commits to it, so it is a one-line category note and the commitment itself comes from elicitation. Produces no specification family: its whole deliverable is evidence a sibling derives from. Runs READ-ONLY: it can fetch and search, and it cannot write, execute or touch the workspace.",
6
+ "functionalTopics": [],
7
+ "preconditions": [
8
+ {
9
+ "kind": "artifact_valid",
10
+ "selector": {
11
+ "family": "vision-brief"
12
+ }
13
+ }
14
+ ],
15
+ "inputs": [
16
+ {
17
+ "family": "vision-brief"
18
+ }
19
+ ],
20
+ "outputs": [],
21
+ "stepOutputs": [
22
+ {
23
+ "ref": "compliance-evidence",
24
+ "condensation": "A brief of at most ~2000 words: each candidate regime by NAME — public law and regulator instruments, never a proprietary certification standard — with the applicability trigger that makes it apply or not apply to THIS product (jurisdiction, data class, sector), the authoritative source consulted (regulator page, official text — never a vendor blog when an official source exists), and an explicit list of what could NOT be established. Every claim carries the URL it came from. Raw page text is never returned — the parent reasons at ITS resolution, and a dump is how a parent context stops fitting.",
25
+ "description": "Source-attributed regime-applicability evidence, condensed for the sibling that derives from it."
26
+ }
27
+ ],
28
+ "method": {
29
+ "guidance": "# Compliance Research — gather the regime evidence\n\n### Target Application Framing\n\nThis knowledge is for configuring behavior in the target application.\nWildo is the framework surface being used, not the product being built.\n\nRead and apply these patterns from the application-developer point of view:\n\n- how the application should configure the relevant app-level file\n- how the application should use framework-provided factories, enums, and helpers\n- how the application should avoid mixing frontend, backend, and infrastructure concerns\n\n\nYou are the **researcher**. Your one job is to find out which regimes plausibly\napply to this product and report it with sources. You do not decide the posture,\nauthor constraints, or touch any file — a sibling does that, from what you bring\nback.\n\n## Why this step exists\n\nThe playbook that records the compliance posture used to derive it from nine\ninterview answers plus what the model remembered about the law. Recollection is\nconfident, fluent, and wrong in exactly the cases that matter here: thresholds\nchange, regimes gain territorial reach, sector rules turn on definitions no one\nremembers precisely — and a wrongly excluded regime is a legal statement about the\nproduct, not a stale price.\n\nThis step exists so the posture rests on pages someone actually read.\n\n## What counts as a regime\n\nA regime is **public law or a regulator's instrument** — a statute, a regulation, a\ndirective as transposed, a supervisory authority's binding guidance. Those are what\nyou research, name and quote: they bind the product whether or not anyone agreed to\nthem, so applicability is a question of fact you can settle from a source.\n\nA **voluntary security certification or attestation framework is not a regime.** It\nbinds only where the creator has committed to it, which makes it an elicited fact\nrather than a researched one — so there is nothing for you to establish and no\nthreshold to test. Report the whole class as one line (\"voluntary\nsecurity-certification and attestation frameworks — commitments, not legal\nobligations; none committed\" or whatever the creator's answers say), and do not name\nan individual proprietary standard or reproduce any of its control identifiers: those\ncatalogues are licensed material, and this evidence store is committed into the\napplication's repository.\n\nThe distinction is not bookkeeping. A list that mixes the two invites a posture to\nread a certification as an obligation the product is failing, which is a false legal\nstatement about the product — the exact defect this lane exists to prevent, arriving\nfrom the other direction.\n\n## What to establish\n\nWork the objective you were given — it was composed for THIS round, and on a later\nround it names what the earlier rounds could not establish. Where it is broad,\nthese are the questions worth answering, per candidate regime:\n\n1. **The applicability trigger, exactly.** What makes it apply — territorial scope,\n data classes, sector, size thresholds, customer commitments — quoted or closely\n paraphrased from an authoritative source, with enough precision that a reader\n can test THIS product against it.\n2. **Whether THIS product trips it**, given the vision's jurisdictions, data\n classes and sector — stated as your reading, distinct from the trigger itself.\n3. **What the regime obliges at posture level.** One or two sentences — the deep\n obligation model is a later concern; the posture needs \"applies / does not\n apply / cannot be established, because …\".\n\n## How to report\n\n- **Authoritative sources first.** A regulator's page or the official text over any\n vendor summary; name the source beside every claim, with its URL.\n- **Say what you could NOT establish.** The gaps aim the next round; hiding them\n produces a posture that reads complete and is not — the exact defect this lane\n replaces.\n- **Never pad.** A regime you found no trigger for is one line, not a paragraph.\n",
30
+ "skillRefs": [
31
+ "journey-status"
32
+ ],
33
+ "contextSources": [],
34
+ "suspendsForClarification": false,
35
+ "briefAdequacyGate": false,
36
+ "briefAdequacyExcludedSources": []
37
+ },
38
+ "executor": "research_agent",
39
+ "approval": "auto",
40
+ "budget": {},
41
+ "onFailure": "keep_and_flag"
42
+ }
@@ -5,10 +5,6 @@
5
5
  "intent": "Produce the market snapshot by gathering EVIDENCE, judging whether it is enough, and deriving the snapshot only once it is. An umbrella: it authors nothing itself — its deliverable is what its children land, and the order between them is decided by their own preconditions rather than declared here.",
6
6
  "functionalTopics": [],
7
7
  "preconditions": [
8
- {
9
- "kind": "phase_is",
10
- "phase": "top_down"
11
- },
12
8
  {
13
9
  "kind": "artifact_valid",
14
10
  "selector": {
@@ -41,7 +37,9 @@
41
37
  "maximumDepth": 1,
42
38
  "maximumDurationMs": 2400000
43
39
  },
44
- "suspendsForClarification": false
40
+ "suspendsForClarification": false,
41
+ "briefAdequacyGate": false,
42
+ "briefAdequacyExcludedSources": []
45
43
  },
46
44
  "approval": "approve_result",
47
45
  "budget": {},
@@ -31,7 +31,9 @@
31
31
  "market-snapshot-validation"
32
32
  ],
33
33
  "contextSources": [],
34
- "suspendsForClarification": false
34
+ "suspendsForClarification": false,
35
+ "briefAdequacyGate": false,
36
+ "briefAdequacyExcludedSources": []
35
37
  },
36
38
  "executor": "companion_agent",
37
39
  "approval": "auto",
@@ -33,7 +33,9 @@
33
33
  "market-alternatives"
34
34
  ],
35
35
  "contextSources": [],
36
- "suspendsForClarification": false
36
+ "suspendsForClarification": false,
37
+ "briefAdequacyGate": false,
38
+ "briefAdequacyExcludedSources": []
37
39
  },
38
40
  "executor": "research_agent",
39
41
  "approval": "auto",
@@ -5,10 +5,6 @@
5
5
  "intent": "Derive a validated market snapshot from the vision and from the EVIDENCE a sibling research step gathered — the artifact-producing child of the `top_down.market-analysis` composite. The provider proposes the market graph; the trusted runtime stamps entity refs/version and keeps provenance, authority and accepted-family mutation outside model output. Its intent used to claim derivation from \"sealed research evidence\" while nothing supplied any; it now runs only once an adequacy gate has read the evidence and returned ADEQUATE.",
6
6
  "functionalTopics": [],
7
7
  "preconditions": [
8
- {
9
- "kind": "phase_is",
10
- "phase": "top_down"
11
- },
12
8
  {
13
9
  "kind": "artifact_valid",
14
10
  "selector": {
@@ -58,7 +54,9 @@
58
54
  "market-snapshot-validation"
59
55
  ],
60
56
  "contextSources": [],
61
- "suspendsForClarification": false
57
+ "suspendsForClarification": false,
58
+ "briefAdequacyGate": false,
59
+ "briefAdequacyExcludedSources": []
62
60
  },
63
61
  "executor": "companion_agent",
64
62
  "approval": "approve_result",
@@ -30,9 +30,12 @@ Two distinct layers must remain visible:
30
30
  structurally and semantically valid on its own. The phase gate is ineligible
31
31
  while `vision-brief`, `market-snapshot`, `requirements`, or `roadmap` is not
32
32
  valid. This floor is non-bypassable.
33
- 2. The compiled Evaluation Protocol interprets cross-family coherence from an
34
- exact sealed evidence set. It cannot make an invalid family valid or replace
35
- missing evidence with an assumption.
33
+ 2. The compiled Evaluation Protocol interprets cross-family coherence from the
34
+ landed input families as they stand in the working tree, beside the COMPUTED
35
+ coherence report (the declared `coherence_report` context section of your brief — the dangling-reference
36
+ and coverage arithmetic is derived exactly and handed in, never re-derived by
37
+ the evaluator). It cannot make an invalid family valid or replace missing evidence
38
+ with an assumption.
36
39
 
37
40
  The canonical criterion instructions, evidence identities, verdict guidance,
38
41
  prompt-assembly strategy, and aggregation precedence live in
@@ -61,7 +64,8 @@ without prose-only inference.
61
64
 
62
65
  ## Evaluation
63
66
 
64
- The evaluator receives the compiled protocol and the exact sealed evidence set.
67
+ The evaluator receives the compiled protocol, the landed input families, and the
68
+ computed coherence report.
65
69
  It must emit one of `PASS`, `CONCERN`, `GAP`, or `INDETERMINATE` for every
66
70
  criterion, cite the evidence used, and state limitations.
67
71
 
@@ -5,10 +5,6 @@
5
5
  "intent": "Assess whether the validated top-down definition set is cross-artifact coherent enough for the current horizon. Deterministic family validation remains a non-bypassable eligibility floor; this playbook contributes a sealed semantic evaluation and never mutates application truth.",
6
6
  "functionalTopics": [],
7
7
  "preconditions": [
8
- {
9
- "kind": "phase_is",
10
- "phase": "top_down"
11
- },
12
8
  {
13
9
  "kind": "artifact_valid",
14
10
  "selector": {
@@ -57,7 +53,7 @@
57
53
  "outputs": [],
58
54
  "stepOutputs": [],
59
55
  "method": {
60
- "guidance": "# Top-Down Phase Gate\n\n### Target Application Framing\n\nThis knowledge is for configuring behavior in the target application.\nWildo is the framework surface being used, not the product being built.\n\nRead and apply these patterns from the application-developer point of view:\n\n- how the application should configure the relevant app-level file\n- how the application should use framework-provided factories, enums, and helpers\n- how the application should avoid mixing frontend, backend, and infrastructure concerns\n\n\n## Purpose\n\nThis playbook assesses whether the accepted top-down definition set is coherent\nenough for the current horizon. It authors no application specification and\ndoes not mutate application truth.\n\nThe assessment is incremental and re-runnable. A later change to market,\napplication-feature, requirements, vision, or roadmap state may invalidate\nearlier evidence and make the decision point eligible again.\n\n## Trust structure\n\nTwo distinct layers must remain visible:\n\n1. The deterministic family-validity floor proves that each required family is\n structurally and semantically valid on its own. The phase gate is ineligible\n while `vision-brief`, `market-snapshot`, `requirements`, or `roadmap` is not\n valid. This floor is non-bypassable.\n2. The compiled Evaluation Protocol interprets cross-family coherence from an\n exact sealed evidence set. It cannot make an invalid family valid or replace\n missing evidence with an assumption.\n\nThe canonical criterion instructions, evidence identities, verdict guidance,\nprompt-assembly strategy, and aggregation precedence live in\n`playbook.definition.ts`. This method explains how to operate the treatment; it\ndoes not duplicate that executable meaning.\n\n## Evidence preparation\n\nUse `journey-status` to identify the current accepted heads for:\n\n- `vision-brief`\n- `market-snapshot`\n- `requirements`\n- `roadmap`\n- `feature-drafts`\n\nProject only the accepted, current-horizon data required by the five protocol\ncriteria. Seal the projected artifacts and their provenance before evaluation.\nIf the evidence projection cannot account for a mandatory requirement, request\nevidence; do not invoke the evaluator on an incomplete set.\n\nBusiness model is not an input to this treatment version and must not be loaded\ninto its knowledge projection. A later treatment version may add a distinct\nbusiness-model criterion when the structured family contracts can support it\nwithout prose-only inference.\n\n## Evaluation\n\nThe evaluator receives the compiled protocol and the exact sealed evidence set.\nIt must emit one of `PASS`, `CONCERN`, `GAP`, or `INDETERMINATE` for every\ncriterion, cite the evidence used, and state limitations.\n\nAggregation uses the protocol’s declared worst-verdict precedence:\n\n`GAP` → `INDETERMINATE` → `CONCERN` → `PASS`\n\n`INDETERMINATE` means that the available evidence is missing, conflicting, or\ntoo uncertain to decide. A transport, model, timeout, or output-validation\nfailure is an evaluator failure and blocks the treatment; it is never rewritten\nas semantic uncertainty.\n\n## Continuation\n\n- `PASS` may continue to the next treatment boundary.\n- `CONCERN` and `GAP` route to revision.\n- `INDETERMINATE` requests additional evidence.\n- a hard-floor failure or evaluator failure blocks.\n\nThis slice deliberately does not activate a runtime evaluator or authority\nmembrane. Human authorization of the top-down → bottom-up transition remains a\nseparate, explicit boundary and must not be inferred from a passing semantic\nassessment. Use `journey-report` to surface the assessment and\n`journey-approve` only when the authority workflow is active.\n\n## Idempotency and bounded work\n\nThe treatment is read-only and safe to repeat against the same accepted heads.\nThe playbook budget limits attempts, evidence requests, and revisions so the\nsystem can reach “good enough for the current horizon” rather than optimizing\nindefinitely. A changed accepted head creates new evidence and therefore a new\nassessment basis.\n",
56
+ "guidance": "# Top-Down Phase Gate\n\n### Target Application Framing\n\nThis knowledge is for configuring behavior in the target application.\nWildo is the framework surface being used, not the product being built.\n\nRead and apply these patterns from the application-developer point of view:\n\n- how the application should configure the relevant app-level file\n- how the application should use framework-provided factories, enums, and helpers\n- how the application should avoid mixing frontend, backend, and infrastructure concerns\n\n\n## Purpose\n\nThis playbook assesses whether the accepted top-down definition set is coherent\nenough for the current horizon. It authors no application specification and\ndoes not mutate application truth.\n\nThe assessment is incremental and re-runnable. A later change to market,\napplication-feature, requirements, vision, or roadmap state may invalidate\nearlier evidence and make the decision point eligible again.\n\n## Trust structure\n\nTwo distinct layers must remain visible:\n\n1. The deterministic family-validity floor proves that each required family is\n structurally and semantically valid on its own. The phase gate is ineligible\n while `vision-brief`, `market-snapshot`, `requirements`, or `roadmap` is not\n valid. This floor is non-bypassable.\n2. The compiled Evaluation Protocol interprets cross-family coherence from the\n landed input families as they stand in the working tree, beside the COMPUTED\n coherence report (the declared `coherence_report` context section of your brief — the dangling-reference\n and coverage arithmetic is derived exactly and handed in, never re-derived by\n the evaluator). It cannot make an invalid family valid or replace missing evidence\n with an assumption.\n\nThe canonical criterion instructions, evidence identities, verdict guidance,\nprompt-assembly strategy, and aggregation precedence live in\n`playbook.definition.ts`. This method explains how to operate the treatment; it\ndoes not duplicate that executable meaning.\n\n## Evidence preparation\n\nUse `journey-status` to identify the current accepted heads for:\n\n- `vision-brief`\n- `market-snapshot`\n- `requirements`\n- `roadmap`\n- `feature-drafts`\n\nProject only the accepted, current-horizon data required by the five protocol\ncriteria. Seal the projected artifacts and their provenance before evaluation.\nIf the evidence projection cannot account for a mandatory requirement, request\nevidence; do not invoke the evaluator on an incomplete set.\n\nBusiness model is not an input to this treatment version and must not be loaded\ninto its knowledge projection. A later treatment version may add a distinct\nbusiness-model criterion when the structured family contracts can support it\nwithout prose-only inference.\n\n## Evaluation\n\nThe evaluator receives the compiled protocol, the landed input families, and the\ncomputed coherence report.\nIt must emit one of `PASS`, `CONCERN`, `GAP`, or `INDETERMINATE` for every\ncriterion, cite the evidence used, and state limitations.\n\nAggregation uses the protocol’s declared worst-verdict precedence:\n\n`GAP` → `INDETERMINATE` → `CONCERN` → `PASS`\n\n`INDETERMINATE` means that the available evidence is missing, conflicting, or\ntoo uncertain to decide. A transport, model, timeout, or output-validation\nfailure is an evaluator failure and blocks the treatment; it is never rewritten\nas semantic uncertainty.\n\n## Continuation\n\n- `PASS` may continue to the next treatment boundary.\n- `CONCERN` and `GAP` route to revision.\n- `INDETERMINATE` requests additional evidence.\n- a hard-floor failure or evaluator failure blocks.\n\nThis slice deliberately does not activate a runtime evaluator or authority\nmembrane. Human authorization of the top-down → bottom-up transition remains a\nseparate, explicit boundary and must not be inferred from a passing semantic\nassessment. Use `journey-report` to surface the assessment and\n`journey-approve` only when the authority workflow is active.\n\n## Idempotency and bounded work\n\nThe treatment is read-only and safe to repeat against the same accepted heads.\nThe playbook budget limits attempts, evidence requests, and revisions so the\nsystem can reach “good enough for the current horizon” rather than optimizing\nindefinitely. A changed accepted head creates new evidence and therefore a new\nassessment basis.\n",
61
57
  "skillRefs": [
62
58
  "journey-status",
63
59
  "journey-report",
@@ -66,7 +62,9 @@
66
62
  "contextSources": [
67
63
  "coherence_report"
68
64
  ],
69
- "suspendsForClarification": false
65
+ "suspendsForClarification": false,
66
+ "briefAdequacyGate": false,
67
+ "briefAdequacyExcludedSources": []
70
68
  },
71
69
  "executor": "companion_agent",
72
70
  "approval": "approve_result",
@@ -44,10 +44,12 @@ The deterministic minimum-content finding and semantic coherence evaluation are
44
44
  separate contributions. Application composition selects exactly one bounded
45
45
  next action from hard floors, scoped risk, maturity, uncertainty, active
46
46
  resource authority and playbook bounds. A model may explain a proposal but it
47
- cannot override a hard floor, choose acceptance authority, or mutate accepted
48
- truth. Under the declared `EXCEPTION_ONLY` supervision posture, an ordinary
49
- eligible result can be policy-accepted only after the acceptance coordinator
50
- revalidates its candidate, evaluation, basis, guards and policy revision.
51
- Protected consequences still request an operator decision. Only the retained
52
- semantic-delta/CAS landing owner may replace accepted application requirements;
53
- later Git materialization remains a separate effect.
47
+ cannot override a hard floor or rewrite accepted truth: it writes files into the
48
+ working tree and nothing more. Acceptance is a human git commit, so a run leaves
49
+ working edits for a person to read and decide on, and protected consequences
50
+ still request an operator decision before the run proceeds.
51
+
52
+ Git is not a later materialization of an acceptance that happened somewhere else
53
+ — it IS the acceptance. Replacing accepted application requirements therefore
54
+ means committing over them, with the commit's author and instant standing as who
55
+ accepted and when.
@@ -5,10 +5,6 @@
5
5
  "intent": "Derive a validated, current-horizon requirements snapshot from accepted vision and market truth. The provider proposes strict semantic content; the trusted runtime stamps family refs/version and keeps provenance, scope, evidence and authority outside model output.",
6
6
  "functionalTopics": [],
7
7
  "preconditions": [
8
- {
9
- "kind": "phase_is",
10
- "phase": "top_down"
11
- },
12
8
  {
13
9
  "kind": "artifact_valid",
14
10
  "selector": {
@@ -40,13 +36,15 @@
40
36
  ],
41
37
  "stepOutputs": [],
42
38
  "method": {
43
- "guidance": "# Requirements definition method\n\nProduce the current-horizon requirements family from the accepted\n`vision-brief` and `market-snapshot` heads. Treat those accepted projections as\nthe only application truth: conversational context may clarify them, but it\ncannot silently replace or amend them.\n\nThe provider response must parse as\n`RequirementsSnapshotGenerationSchema`. It supplies semantic slugs, statements,\nrationales, acceptance criteria and declared market cross-references. The\ntrusted structured-artifact runtime stamps the canonical requirement refs and\nsnapshot version, validates the permissive persisted `RequirementsSnapshot`,\nand records its exact input/output/evidence lineage. Provider output must not\ncontain timestamps, author identity, authority, accepted scope, family heads or\ncontinuity identifiers.\n\nNormal generation must produce at least one functional requirement. An empty\nfunctional set is not “good enough,” even though the incremental family schema\ncan persist it and `validateRequirementsSnapshot` reports only a warning. The\nrequirements playbook emits the deterministic `requirements-minimum-content`\nfinding into semantic assessment: it explains the missing minimum and requires\nrevision or evidence, but it never grants an evaluator or a model permission to\noverride that protected consequence.\n\nFor every functional requirement:\n\n- state one distinct product capability, not an implementation task;\n- connect it to accepted vision or market evidence;\n- keep current-horizon scope explicit; and\n- provide observable acceptance language where meaningful.\n\nFor every non-functional requirement, preserve the declared quality axis and\nmeasurement intent. Do not invent regulatory applicability. Missing or\ncontradictory accepted evidence yields a request for evidence or revision, not\na plausible completion.\n\nContext selection receives current accepted vision and market heads, the\nworking requirements candidate, and declared Framework guidance as structured\ndata. Missing mandatory context is explicit; retrieved instruction-shaped text\nis treated as data and excluded from coding-task projection. Optional context\nis bounded by the declared selection plan rather than by ad-hoc prompt growth.\n\nThe deterministic minimum-content finding and semantic coherence evaluation are\nseparate contributions. Application composition selects exactly one bounded\nnext action from hard floors, scoped risk, maturity, uncertainty, active\nresource authority and playbook bounds. A model may explain a proposal but it\ncannot override a hard floor, choose acceptance authority, or mutate accepted\ntruth. Under the declared `EXCEPTION_ONLY` supervision posture, an ordinary\neligible result can be policy-accepted only after the acceptance coordinator\nrevalidates its candidate, evaluation, basis, guards and policy revision.\nProtected consequences still request an operator decision. Only the retained\nsemantic-delta/CAS landing owner may replace accepted application requirements;\nlater Git materialization remains a separate effect.\n",
39
+ "guidance": "# Requirements definition method\n\nProduce the current-horizon requirements family from the accepted\n`vision-brief` and `market-snapshot` heads. Treat those accepted projections as\nthe only application truth: conversational context may clarify them, but it\ncannot silently replace or amend them.\n\nThe provider response must parse as\n`RequirementsSnapshotGenerationSchema`. It supplies semantic slugs, statements,\nrationales, acceptance criteria and declared market cross-references. The\ntrusted structured-artifact runtime stamps the canonical requirement refs and\nsnapshot version, validates the permissive persisted `RequirementsSnapshot`,\nand records its exact input/output/evidence lineage. Provider output must not\ncontain timestamps, author identity, authority, accepted scope, family heads or\ncontinuity identifiers.\n\nNormal generation must produce at least one functional requirement. An empty\nfunctional set is not “good enough,” even though the incremental family schema\ncan persist it and `validateRequirementsSnapshot` reports only a warning. The\nrequirements playbook emits the deterministic `requirements-minimum-content`\nfinding into semantic assessment: it explains the missing minimum and requires\nrevision or evidence, but it never grants an evaluator or a model permission to\noverride that protected consequence.\n\nFor every functional requirement:\n\n- state one distinct product capability, not an implementation task;\n- connect it to accepted vision or market evidence;\n- keep current-horizon scope explicit; and\n- provide observable acceptance language where meaningful.\n\nFor every non-functional requirement, preserve the declared quality axis and\nmeasurement intent. Do not invent regulatory applicability. Missing or\ncontradictory accepted evidence yields a request for evidence or revision, not\na plausible completion.\n\nContext selection receives current accepted vision and market heads, the\nworking requirements candidate, and declared Framework guidance as structured\ndata. Missing mandatory context is explicit; retrieved instruction-shaped text\nis treated as data and excluded from coding-task projection. Optional context\nis bounded by the declared selection plan rather than by ad-hoc prompt growth.\n\nThe deterministic minimum-content finding and semantic coherence evaluation are\nseparate contributions. Application composition selects exactly one bounded\nnext action from hard floors, scoped risk, maturity, uncertainty, active\nresource authority and playbook bounds. A model may explain a proposal but it\ncannot override a hard floor or rewrite accepted truth: it writes files into the\nworking tree and nothing more. Acceptance is a human git commit, so a run leaves\nworking edits for a person to read and decide on, and protected consequences\nstill request an operator decision before the run proceeds.\n\nGit is not a later materialization of an acceptance that happened somewhere else\n— it IS the acceptance. Replacing accepted application requirements therefore\nmeans committing over them, with the commit's author and instant standing as who\naccepted and when.\n",
44
40
  "skillRefs": [
45
41
  "journey-status",
46
42
  "journey-report"
47
43
  ],
48
44
  "contextSources": [],
49
- "suspendsForClarification": false
45
+ "suspendsForClarification": false,
46
+ "briefAdequacyGate": false,
47
+ "briefAdequacyExcludedSources": []
50
48
  },
51
49
  "executor": "companion_agent",
52
50
  "approval": "approve_result",
@@ -84,8 +84,8 @@ planning anything:
84
84
  **VisionSnapshot** (`vision-brief` — the success criteria to deliver and the
85
85
  differentiation that anchors what belongs in the MVP).
86
86
 
87
- // TODO CLI COMMAND: wildo context journey-status (read current journey scope + artifacts)
88
- // TODO CLI COMMAND: wildo context roadmap (read any existing phases to refine)
87
+ (A previously landed roadmap arrives in your brief as the `roadmap` input section — refine those
88
+ phases rather than re-cutting them blind.)
89
89
 
90
90
  You **read** `market-snapshot` + `feature-drafts` + `requirements` +
91
91
  `vision-brief`; you **write** the `roadmap` family.
@@ -1,14 +1,10 @@
1
1
  {
2
2
  "ref": "top_down.scope-roadmap",
3
- "version": "0.1.0",
3
+ "version": "0.1.1",
4
4
  "phase": "top_down",
5
5
  "intent": "Turn the accepted capability map into a non-vacuous current-horizon RoadmapSnapshot. Phases explicitly sequence and group accepted requirements and vision success criteria without pretending that FIT is build-order authority or inventing unsupported delivery commitments.",
6
6
  "functionalTopics": [],
7
7
  "preconditions": [
8
- {
9
- "kind": "phase_is",
10
- "phase": "top_down"
11
- },
12
8
  {
13
9
  "kind": "artifact_valid",
14
10
  "selector": {
@@ -44,6 +40,9 @@
44
40
  {
45
41
  "family": "requirements"
46
42
  },
43
+ {
44
+ "family": "roadmap"
45
+ },
47
46
  {
48
47
  "family": "vision-brief"
49
48
  }
@@ -57,7 +56,7 @@
57
56
  ],
58
57
  "stepOutputs": [],
59
58
  "method": {
60
- "guidance": "# Scope & Roadmap — Capability Map → Release Plan Method (v1 scaffold)\n\n### Target Application Framing\n\nThis knowledge is for configuring behavior in the target application.\nWildo is the framework surface being used, not the product being built.\n\nRead and apply these patterns from the application-developer point of view:\n\n- how the application should configure the relevant app-level file\n- how the application should use framework-provided factories, enums, and helpers\n- how the application should avoid mixing frontend, backend, and infrastructure concerns\n\n\n> **This is a v1 scaffold methodology.** It is a working first cut meant to be\n> iterated. The phasing order, the MVP-first rubric, and the in/out-of-scope\n> convention below are deliberately explicit so they can be tuned against real\n> runs — treat them as a starting rubric, not a frozen contract. The\n> *analytical depth* of release planning (typed schedules, dependency graphs,\n> capacity/estimation, defended prioritization) is intentionally shallow here\n> and is deepened in a later phase; do not perfect it in this playbook.\n\n## What this playbook does\n\nYou are the **release planner**. This playbook turns *product capability* (the\nfeatures the capability map identified and the requirements behind them) into a\n**release plan**: an ordered set of roadmap **phases** (MVP → Beta → GA →\nGrowth) that declare **what gets delivered when**. Capability mapping already\nproduced a **MarketSnapshot** whose benefits are wired to delivering features,\nplus **feature-drafts** and a set of **requirements**. Your one job is to answer,\nfor the whole product: **in what order do we deliver this, and what is in vs out\nof scope for each release?** — and record the answer as a `RoadmapSnapshot`.\n\nYou produce ONE artifact for this journey: the **roadmap** family\n(`RoadmapSnapshot`), a single ordered `phases` collection.\n\nThis is a **focused planning pass**, distinct from the capability map (which\nmade each benefit deliverable and drafted the features) and from the phase-gate\n(which decides whether the top-down phase is complete). Where the capability map\nsaid *what the product must do*, you say *in what order it ships*.\n\n## The roadmap is requirement/success-criterion-addressed, NOT feature-addressed\n\nRead this before you start — it is the one structural fact that shapes the whole\nmethod. A roadmap phase declares what it delivers through **two flat cross-ref\nlists**:\n\n- **`deliversRequirementRefs`** — the requirement refs (from the `requirements`\n family) this phase delivers. Primarily **functional-requirement** refs\n (`functional-requirement-…`) — the case the `_core` phase contract narrates —\n but a phase that lands a quality attribute may also deliver a **non-functional**\n ref (`nfr-…`); the catalog resolves both against one combined requirement-ref\n pool, so an NFR ref is valid here (see the NFR-phasing boundary note).\n- **`deliversSuccessCriterionRefs`** — the vision success-criterion refs (from\n the `vision-brief` family) this phase is expected to satisfy.\n\nA phase does **not** name feature refs. So the capability map's **features reach\nthe roadmap indirectly**: feature → benefit → requirement → phase. You slot\n**requirements and success criteria** into phases; you use the capability map\n(which feature delivers which benefit, and the FIT/Kano read on the feature set)\nto **judge which requirements are MVP-critical vs later**. The feature →\nrequirement trace is **prose-only today** (a requirement cross-refs ICPs and jobs\nbut not market pains/gains, and there is no typed feature → requirement edge —\nthis is the filed depth-phase gap; see the scaffold-depth boundary). At scaffold\ndepth you bridge that trace by reading the capability map and the requirements\ntogether and using judgement — do **not** try to make it deterministic here.\n\n## Read state first — the four families you read, the one you write\n\nThis playbook is **idempotent** and, on a re-run, a *refinement*: a MarketSnapshot\nmust already exist (the `artifact_exists` precondition enforces the capability map\nis in place — you cannot plan releases before you know the capabilities). Before\nplanning anything:\n\n1. Read the current journey state and scope through the `journey-status` skill\n (it owns the read; do not re-derive it here). Confirm a `market-snapshot`\n exists and check whether a `roadmap` already exists from an earlier run.\n2. Read the capability-mapped **MarketSnapshot** (benefits + their\n `deliveredByFeatureRefs` + FIT/Kano signal), the **feature-drafts** (the prose\n capability contracts — what each delivering feature actually is, so you can\n weigh how much of the product a requirement pulls in and whether it is\n MVP-critical), the **requirements** (`requirements` family — the\n functional/non-functional requirements to slot into phases), and the\n **VisionSnapshot** (`vision-brief` — the success criteria to deliver and the\n differentiation that anchors what belongs in the MVP).\n\n// TODO CLI COMMAND: wildo context journey-status (read current journey scope + artifacts)\n// TODO CLI COMMAND: wildo context roadmap (read any existing phases to refine)\n\nYou **read** `market-snapshot` + `feature-drafts` + `requirements` +\n`vision-brief`; you **write** the `roadmap` family.\n\n## The phasing loop (build the ordered releases)\n\n1. **Establish the MVP cut first.** The first phase is the smallest coherent\n release that makes the product's core promise true. Pull in the requirements\n whose delivering features are **table-stakes** for the primary ICP — the\n capability map's FIT/Kano read surfaces these (a `THRESHOLD` benefit for a\n `CORE`/`EXPANSION` ICP is MVP-critical). A requirement whose benefit is a\n `DELIGHTER`, or serves a non-primary segment, is a candidate for a later\n phase, not the MVP. MVP-first is the scaffold rubric: deliver the differentiated\n core early, defer the upside.\n2. **Group the remaining work into ordered later phases.** Beta / GA / Growth\n stages each pull in the next coherent slice of requirements and success\n criteria. Group by *theme* (a phase has a narrative headline — \"Trust &\n safety\", \"Onboarding excellence\"), not by arbitrary size.\n3. **Assign each phase its ordering + coarse tags.**\n - **`sequence`** — the authoritative ordering key (ascending integer).\n Consumers sort by `sequence`, never by array position. Author with **gaps**\n (10, 20, 30) so a later phase can be inserted without renumbering. Every\n phase must have a **distinct** `sequence` (a shared sequence is a\n `DUPLICATE_SEQUENCE` WARN — the order becomes ambiguous).\n - **`horizon`** — coarse `NOW` / `NEXT` / `LATER` bucket (`RoadmapPhaseHorizon`).\n - **`kind`** — coarse lifecycle stage `DISCOVERY` / `MVP` / `BETA` / `GA` /\n `GROWTH` / `OTHER` (`RoadmapPhaseKind`).\n - **`title`** (required), plus optional `theme` (narrative angle),\n `objective` (the concrete outcome in prose), and `targetWindow` (WHEN, as\n **free prose** — \"Q3 2026\", \"first 3 months post-launch\"; a typed schedule\n is a later-phase concern).\n4. **Wire what each phase delivers.** Populate `deliversRequirementRefs` with the\n requirement refs (functional, and any non-functional quality the phase lands)\n the phase delivers, and `deliversSuccessCriterionRefs` with the vision\n success-criterion refs it satisfies. **Every requirement and every success\n criterion should land in at most one phase** — the earliest that delivers it.\n Two distinct duplication cases exist, and the gate catches only one:\n - **Same ref twice on ONE phase's list** → a `DUPLICATE_CROSS_REF` WARN. The\n gate catches this (each of the two lists is its own namespace, so the same\n string appearing once in each list is NOT flagged).\n - **Same ref on TWO different phases** → the gate does **NOT** catch this — the\n validator's duplicate check is per-list, per-phase, never cross-phase. It is\n silently wrong: coverage double-counts the ref and \"which release first\n delivers it\" becomes ambiguous. Keeping a ref in exactly one phase is an\n **authoring discipline**, not a gate-enforced invariant.\n5. **Sanity-check coverage.** By the last phase, every requirement whose benefit\n the product actually promises, and every vision success criterion, should be\n delivered by some phase. A requirement no phase delivers is out-of-scope of the\n *entire roadmap* — that is a real decision (it says \"the product will not do\n this\"), so make it deliberately and narrate it, do not leave it by accident.\n\n## In-scope vs out-of-scope per release (the scope decision)\n\nThe roadmap `_core` has **no typed out-of-scope axis** — a phase declares only\nwhat it **delivers**. So scope is modelled as follows at scaffold depth:\n\n- **In-scope of a release = its `delivers*` refs.** The `deliversRequirementRefs`\n + `deliversSuccessCriterionRefs` a phase carries ARE its declared, gate-checked\n in-scope. This is explicit and typed.\n- **Out-of-scope of a release is implicit + partly derivable.** Everything NOT\n yet delivered by a phase at or before this one is out-of-scope of that release\n (cumulative — a ref delivered by an earlier phase stays in-scope in every later\n one). Within that out-of-scope set, two sub-cases ARE derivable from the roadmap\n alone: *deferred* (some later phase delivers the ref → planned, just not yet)\n vs *never* (NO phase anywhere delivers the ref → out of the product).\n- **What is NOT derivable — and therefore needs prose.** The roadmap cannot\n distinguish a *deliberately excluded* requirement (\"we chose not to build SSO\")\n from an *accidentally dropped* one (\"we forgot to slot it\"): both surface\n identically — a requirement authored in the `requirements` family that no phase\n delivers. That single ambiguity is the entire cost of Option A. So record the\n **rationale** for every deliberate \"never\" in the phase `objective` / `theme`\n **prose** (\"Out of scope for MVP: bulk import, SSO — deferred to GA\"; \"Not in\n the product: on-prem hosting — SaaS-only by design\"), and run the coverage\n sanity-check (step 5 above) so an accidental drop is caught rather than shipped\n as a phantom exclusion.\n\nRejected alternatives:\n\n- A **typed out-of-scope / explicitly-deferred axis** on a phase (a\n `defersRequirementRefs` list, or an `outOfScope` block). It is the real fix for\n the ambiguity above — it makes \"deliberately never\" machine-readable — but it is\n a `_core/roadmap` reshape, filed as a depth-phase item, NOT built here. Do\n **not** reshape the roadmap contract to add it.\n- A **phantom \"out-of-scope\" phase** — a phase whose `deliversRequirementRefs`\n lists the *excluded* requirements to fake a machine-readable exclusion set. Do\n **NOT** do this. A phase *delivers* its refs by definition; listing excluded\n refs there inverts the semantics and **corrupts every downstream coverage\n count** (those refs would read as delivered, and the phase-gate / refinement\n playbooks would treat excluded work as shipped). This is exactly the kind of\n \"trick\" the framework mindset forbids — the excluded set stays in prose until\n the typed axis lands, never by abusing `delivers*`.\n\n## Refine, don't re-author\n\n- **Keep what still holds.** On a re-run, do not renumber or re-theme phases that\n are already correct. Only add a phase, move a requirement to a different phase,\n or correct a delivers-list where the plan genuinely changed.\n- **Reuse existing phase refs.** A phase that already exists keeps its\n `roadmap-phase-<slug>` ref — never mint a second ref for the same release. Refs\n are unique (a `DUPLICATE_REF` is an ERROR that silently drops a phase).\n- **Never orphan a deliver-ref.** A phase's `deliversRequirementRefs` /\n `deliversSuccessCriterionRefs` must name refs that exist in the requirements /\n vision families. These cross-family refs are resolution-checked **fail-soft**\n (the check is skipped while the families are decoupled and no ref set is\n supplied), so honest planning must not name a requirement or success criterion\n that does not exist in the first place.\n- **Each ref in at most one phase.** A requirement/success-criterion is delivered\n by the earliest phase that delivers it; do not re-list it in later phases.\n\n## Quality gate\n\nThe RoadmapSnapshot is checked by the deterministic `roadmap-valid` gate\n(`validateRoadmapSnapshot`) before anything downstream consumes it. The invariants\nit enforces (verified issue catalogue):\n\n- **`INVALID_REF_FORMAT` (ERROR)** — a phase ref is not kebab-case with the\n `roadmap-phase-` prefix.\n- **`DUPLICATE_REF` (ERROR)** — two phases share a ref; the collection is\n addressed by ref, so a duplicate silently drops a phase.\n- **`DUPLICATE_SEQUENCE` (WARN)** — two phases share a `sequence`; the order\n between them is ambiguous (no data lost, so a WARN). Give each phase a distinct\n `sequence`.\n- **`DUPLICATE_CROSS_REF` (WARN)** — the same deliver-ref is listed more than\n once on ONE phase's deliver-list (each of the two lists is its own namespace, so\n the same ref appearing once in each list is NOT a duplicate). Remove the repeat.\n- **`NO_PHASES` (WARN)** — the roadmap plans no phases; it describes no delivery.\n- **`STALE_AUTHORED` (WARN)** — a phase's `authored.lastAuthoredAt` is older than\n its `reviewCadence`.\n- **`UNKNOWN_REQUIREMENT_REF` / `UNKNOWN_SUCCESS_CRITERION_REF` (ERROR, fail-soft)**\n — a `deliversRequirementRefs` / `deliversSuccessCriterionRefs` entry does not\n resolve against a supplied ref set. These are **SKIPPED (fail-soft) when no ref\n set is passed** — the roadmap family is decoupled from the requirements/vision\n families and the catalog owns cross-family resolution, so authoring a roadmap\n before those registries are fully wired does NOT crash-loop. Even so, only name\n refs that really exist (see \"Never orphan a deliver-ref\").\n\n## Scaffold-depth boundary (deferred to a later depth phase)\n\nKeep this playbook at scaffold depth. Explicitly **out of scope here**:\n\n- **Typed timeline / scheduling.** `targetWindow` is free prose and ordering is a\n coarse `sequence` integer. Start/end dates, date math, and calendar alignment\n are a later-phase concern — a phase says \"first 3 months post-launch\" as prose\n today, not a typed interval. Do NOT build a schedule model.\n- **Inter-phase dependency graph.** Phases are an ordered list, not a DAG. Edges\n between phases (phase B needs phase A's delivery) are a later-phase concern.\n- **Capacity / velocity / estimation and prioritization frameworks.** A phase is a\n themed bucket of deliver-refs, not an estimated, capacity-fitted, RICE/WSJF/MoSCoW-\n ranked plan. If you catch yourself estimating effort or building a prioritization\n model, **stop** — that is depth-phase work.\n- **A typed out-of-scope / explicitly-deferred axis** (see the scope decision\n above) — a `_core/roadmap` reshape, filed as a depth-phase item, not built here.\n- **Requirement priority + inter-requirement dependency.** A requirement has no\n MoSCoW/priority axis and no \"requirement A needs B\" edge; only the roadmap\n `sequence` approximates ordering. Do NOT reshape the requirements contract to add\n them — it is a filed depth-phase question.\n- **A dedicated non-functional-requirement delivery axis.** `deliversRequirementRefs`\n today carries functional AND non-functional refs against one combined pool. Some\n NFRs are cross-cutting quality attributes (accessibility, latency) that no single\n phase \"delivers\" — whether NFRs deserve their own phase-delivery axis (or a\n per-phase quality bar) rather than sharing `deliversRequirementRefs` is a\n depth-phase question. At scaffold depth, slot a phase-specific NFR into\n `deliversRequirementRefs` and leave universal NFRs unphased; do NOT reshape\n `_core` for a separate axis here.\n- **Per-phase ICP/segment targeting.** A phase cannot declare the primary\n ICP/segment it serves; that is a filed depth-phase gap. Note it in prose if it\n matters; do NOT reshape `_core` to add a typed edge.\n- **Deterministic feature → phase traceability.** Because a requirement carries no\n typed market pain/gain edge (and no typed feature → requirement edge), the\n feature → benefit → requirement → phase trace is prose-only. Whether the\n requirements family should gain a pain/gain edge is a filed depth-phase question —\n do NOT reshape the requirements or roadmap contract here.\n\nIf you find yourself perfecting any of the above, stop and leave the slice at\nscaffold depth — depth is a dedicated later pass done once, holistically.\n\n## What you emit\n\nOne artifact for this journey: the **roadmap** at\n`specifications/src/roadmap/index.ts` — a `RoadmapSnapshot` whose `phases`\nare the ordered, MVP-first releases, each with a distinct `sequence`, a coarse\n`horizon` + `kind`, and its `deliversRequirementRefs` / `deliversSuccessCriterionRefs`\nwired — passing the `roadmap-valid` gate.\n\nThe creator reviews the result (approval posture `approve_result`) before it is\naccepted as the journey's roadmap.\n\n## References\n\n- `.claude/skills/journey-status/SKILL.md` — read the current journey state and\n scope, and confirm a `market-snapshot` (the capability map) already exists to\n plan releases from.\n",
59
+ "guidance": "# Scope & Roadmap — Capability Map → Release Plan Method (v1 scaffold)\n\n### Target Application Framing\n\nThis knowledge is for configuring behavior in the target application.\nWildo is the framework surface being used, not the product being built.\n\nRead and apply these patterns from the application-developer point of view:\n\n- how the application should configure the relevant app-level file\n- how the application should use framework-provided factories, enums, and helpers\n- how the application should avoid mixing frontend, backend, and infrastructure concerns\n\n\n> **This is a v1 scaffold methodology.** It is a working first cut meant to be\n> iterated. The phasing order, the MVP-first rubric, and the in/out-of-scope\n> convention below are deliberately explicit so they can be tuned against real\n> runs — treat them as a starting rubric, not a frozen contract. The\n> *analytical depth* of release planning (typed schedules, dependency graphs,\n> capacity/estimation, defended prioritization) is intentionally shallow here\n> and is deepened in a later phase; do not perfect it in this playbook.\n\n## What this playbook does\n\nYou are the **release planner**. This playbook turns *product capability* (the\nfeatures the capability map identified and the requirements behind them) into a\n**release plan**: an ordered set of roadmap **phases** (MVP → Beta → GA →\nGrowth) that declare **what gets delivered when**. Capability mapping already\nproduced a **MarketSnapshot** whose benefits are wired to delivering features,\nplus **feature-drafts** and a set of **requirements**. Your one job is to answer,\nfor the whole product: **in what order do we deliver this, and what is in vs out\nof scope for each release?** — and record the answer as a `RoadmapSnapshot`.\n\nYou produce ONE artifact for this journey: the **roadmap** family\n(`RoadmapSnapshot`), a single ordered `phases` collection.\n\nThis is a **focused planning pass**, distinct from the capability map (which\nmade each benefit deliverable and drafted the features) and from the phase-gate\n(which decides whether the top-down phase is complete). Where the capability map\nsaid *what the product must do*, you say *in what order it ships*.\n\n## The roadmap is requirement/success-criterion-addressed, NOT feature-addressed\n\nRead this before you start — it is the one structural fact that shapes the whole\nmethod. A roadmap phase declares what it delivers through **two flat cross-ref\nlists**:\n\n- **`deliversRequirementRefs`** — the requirement refs (from the `requirements`\n family) this phase delivers. Primarily **functional-requirement** refs\n (`functional-requirement-…`) — the case the `_core` phase contract narrates —\n but a phase that lands a quality attribute may also deliver a **non-functional**\n ref (`nfr-…`); the catalog resolves both against one combined requirement-ref\n pool, so an NFR ref is valid here (see the NFR-phasing boundary note).\n- **`deliversSuccessCriterionRefs`** — the vision success-criterion refs (from\n the `vision-brief` family) this phase is expected to satisfy.\n\nA phase does **not** name feature refs. So the capability map's **features reach\nthe roadmap indirectly**: feature → benefit → requirement → phase. You slot\n**requirements and success criteria** into phases; you use the capability map\n(which feature delivers which benefit, and the FIT/Kano read on the feature set)\nto **judge which requirements are MVP-critical vs later**. The feature →\nrequirement trace is **prose-only today** (a requirement cross-refs ICPs and jobs\nbut not market pains/gains, and there is no typed feature → requirement edge —\nthis is the filed depth-phase gap; see the scaffold-depth boundary). At scaffold\ndepth you bridge that trace by reading the capability map and the requirements\ntogether and using judgement — do **not** try to make it deterministic here.\n\n## Read state first — the four families you read, the one you write\n\nThis playbook is **idempotent** and, on a re-run, a *refinement*: a MarketSnapshot\nmust already exist (the `artifact_exists` precondition enforces the capability map\nis in place — you cannot plan releases before you know the capabilities). Before\nplanning anything:\n\n1. Read the current journey state and scope through the `journey-status` skill\n (it owns the read; do not re-derive it here). Confirm a `market-snapshot`\n exists and check whether a `roadmap` already exists from an earlier run.\n2. Read the capability-mapped **MarketSnapshot** (benefits + their\n `deliveredByFeatureRefs` + FIT/Kano signal), the **feature-drafts** (the prose\n capability contracts — what each delivering feature actually is, so you can\n weigh how much of the product a requirement pulls in and whether it is\n MVP-critical), the **requirements** (`requirements` family — the\n functional/non-functional requirements to slot into phases), and the\n **VisionSnapshot** (`vision-brief` — the success criteria to deliver and the\n differentiation that anchors what belongs in the MVP).\n\n(A previously landed roadmap arrives in your brief as the `roadmap` input section — refine those\nphases rather than re-cutting them blind.)\n\nYou **read** `market-snapshot` + `feature-drafts` + `requirements` +\n`vision-brief`; you **write** the `roadmap` family.\n\n## The phasing loop (build the ordered releases)\n\n1. **Establish the MVP cut first.** The first phase is the smallest coherent\n release that makes the product's core promise true. Pull in the requirements\n whose delivering features are **table-stakes** for the primary ICP — the\n capability map's FIT/Kano read surfaces these (a `THRESHOLD` benefit for a\n `CORE`/`EXPANSION` ICP is MVP-critical). A requirement whose benefit is a\n `DELIGHTER`, or serves a non-primary segment, is a candidate for a later\n phase, not the MVP. MVP-first is the scaffold rubric: deliver the differentiated\n core early, defer the upside.\n2. **Group the remaining work into ordered later phases.** Beta / GA / Growth\n stages each pull in the next coherent slice of requirements and success\n criteria. Group by *theme* (a phase has a narrative headline — \"Trust &\n safety\", \"Onboarding excellence\"), not by arbitrary size.\n3. **Assign each phase its ordering + coarse tags.**\n - **`sequence`** — the authoritative ordering key (ascending integer).\n Consumers sort by `sequence`, never by array position. Author with **gaps**\n (10, 20, 30) so a later phase can be inserted without renumbering. Every\n phase must have a **distinct** `sequence` (a shared sequence is a\n `DUPLICATE_SEQUENCE` WARN — the order becomes ambiguous).\n - **`horizon`** — coarse `NOW` / `NEXT` / `LATER` bucket (`RoadmapPhaseHorizon`).\n - **`kind`** — coarse lifecycle stage `DISCOVERY` / `MVP` / `BETA` / `GA` /\n `GROWTH` / `OTHER` (`RoadmapPhaseKind`).\n - **`title`** (required), plus optional `theme` (narrative angle),\n `objective` (the concrete outcome in prose), and `targetWindow` (WHEN, as\n **free prose** — \"Q3 2026\", \"first 3 months post-launch\"; a typed schedule\n is a later-phase concern).\n4. **Wire what each phase delivers.** Populate `deliversRequirementRefs` with the\n requirement refs (functional, and any non-functional quality the phase lands)\n the phase delivers, and `deliversSuccessCriterionRefs` with the vision\n success-criterion refs it satisfies. **Every requirement and every success\n criterion should land in at most one phase** — the earliest that delivers it.\n Two distinct duplication cases exist, and the gate catches only one:\n - **Same ref twice on ONE phase's list** → a `DUPLICATE_CROSS_REF` WARN. The\n gate catches this (each of the two lists is its own namespace, so the same\n string appearing once in each list is NOT flagged).\n - **Same ref on TWO different phases** → the gate does **NOT** catch this — the\n validator's duplicate check is per-list, per-phase, never cross-phase. It is\n silently wrong: coverage double-counts the ref and \"which release first\n delivers it\" becomes ambiguous. Keeping a ref in exactly one phase is an\n **authoring discipline**, not a gate-enforced invariant.\n5. **Sanity-check coverage.** By the last phase, every requirement whose benefit\n the product actually promises, and every vision success criterion, should be\n delivered by some phase. A requirement no phase delivers is out-of-scope of the\n *entire roadmap* — that is a real decision (it says \"the product will not do\n this\"), so make it deliberately and narrate it, do not leave it by accident.\n\n## In-scope vs out-of-scope per release (the scope decision)\n\nThe roadmap `_core` has **no typed out-of-scope axis** — a phase declares only\nwhat it **delivers**. So scope is modelled as follows at scaffold depth:\n\n- **In-scope of a release = its `delivers*` refs.** The `deliversRequirementRefs`\n + `deliversSuccessCriterionRefs` a phase carries ARE its declared, gate-checked\n in-scope. This is explicit and typed.\n- **Out-of-scope of a release is implicit + partly derivable.** Everything NOT\n yet delivered by a phase at or before this one is out-of-scope of that release\n (cumulative — a ref delivered by an earlier phase stays in-scope in every later\n one). Within that out-of-scope set, two sub-cases ARE derivable from the roadmap\n alone: *deferred* (some later phase delivers the ref → planned, just not yet)\n vs *never* (NO phase anywhere delivers the ref → out of the product).\n- **What is NOT derivable — and therefore needs prose.** The roadmap cannot\n distinguish a *deliberately excluded* requirement (\"we chose not to build SSO\")\n from an *accidentally dropped* one (\"we forgot to slot it\"): both surface\n identically — a requirement authored in the `requirements` family that no phase\n delivers. That single ambiguity is the entire cost of Option A. So record the\n **rationale** for every deliberate \"never\" in the phase `objective` / `theme`\n **prose** (\"Out of scope for MVP: bulk import, SSO — deferred to GA\"; \"Not in\n the product: on-prem hosting — SaaS-only by design\"), and run the coverage\n sanity-check (step 5 above) so an accidental drop is caught rather than shipped\n as a phantom exclusion.\n\nRejected alternatives:\n\n- A **typed out-of-scope / explicitly-deferred axis** on a phase (a\n `defersRequirementRefs` list, or an `outOfScope` block). It is the real fix for\n the ambiguity above — it makes \"deliberately never\" machine-readable — but it is\n a `_core/roadmap` reshape, filed as a depth-phase item, NOT built here. Do\n **not** reshape the roadmap contract to add it.\n- A **phantom \"out-of-scope\" phase** — a phase whose `deliversRequirementRefs`\n lists the *excluded* requirements to fake a machine-readable exclusion set. Do\n **NOT** do this. A phase *delivers* its refs by definition; listing excluded\n refs there inverts the semantics and **corrupts every downstream coverage\n count** (those refs would read as delivered, and the phase-gate / refinement\n playbooks would treat excluded work as shipped). This is exactly the kind of\n \"trick\" the framework mindset forbids — the excluded set stays in prose until\n the typed axis lands, never by abusing `delivers*`.\n\n## Refine, don't re-author\n\n- **Keep what still holds.** On a re-run, do not renumber or re-theme phases that\n are already correct. Only add a phase, move a requirement to a different phase,\n or correct a delivers-list where the plan genuinely changed.\n- **Reuse existing phase refs.** A phase that already exists keeps its\n `roadmap-phase-<slug>` ref — never mint a second ref for the same release. Refs\n are unique (a `DUPLICATE_REF` is an ERROR that silently drops a phase).\n- **Never orphan a deliver-ref.** A phase's `deliversRequirementRefs` /\n `deliversSuccessCriterionRefs` must name refs that exist in the requirements /\n vision families. These cross-family refs are resolution-checked **fail-soft**\n (the check is skipped while the families are decoupled and no ref set is\n supplied), so honest planning must not name a requirement or success criterion\n that does not exist in the first place.\n- **Each ref in at most one phase.** A requirement/success-criterion is delivered\n by the earliest phase that delivers it; do not re-list it in later phases.\n\n## Quality gate\n\nThe RoadmapSnapshot is checked by the deterministic `roadmap-valid` gate\n(`validateRoadmapSnapshot`) before anything downstream consumes it. The invariants\nit enforces (verified issue catalogue):\n\n- **`INVALID_REF_FORMAT` (ERROR)** — a phase ref is not kebab-case with the\n `roadmap-phase-` prefix.\n- **`DUPLICATE_REF` (ERROR)** — two phases share a ref; the collection is\n addressed by ref, so a duplicate silently drops a phase.\n- **`DUPLICATE_SEQUENCE` (WARN)** — two phases share a `sequence`; the order\n between them is ambiguous (no data lost, so a WARN). Give each phase a distinct\n `sequence`.\n- **`DUPLICATE_CROSS_REF` (WARN)** — the same deliver-ref is listed more than\n once on ONE phase's deliver-list (each of the two lists is its own namespace, so\n the same ref appearing once in each list is NOT a duplicate). Remove the repeat.\n- **`NO_PHASES` (WARN)** — the roadmap plans no phases; it describes no delivery.\n- **`STALE_AUTHORED` (WARN)** — a phase's `authored.lastAuthoredAt` is older than\n its `reviewCadence`.\n- **`UNKNOWN_REQUIREMENT_REF` / `UNKNOWN_SUCCESS_CRITERION_REF` (ERROR, fail-soft)**\n — a `deliversRequirementRefs` / `deliversSuccessCriterionRefs` entry does not\n resolve against a supplied ref set. These are **SKIPPED (fail-soft) when no ref\n set is passed** — the roadmap family is decoupled from the requirements/vision\n families and the catalog owns cross-family resolution, so authoring a roadmap\n before those registries are fully wired does NOT crash-loop. Even so, only name\n refs that really exist (see \"Never orphan a deliver-ref\").\n\n## Scaffold-depth boundary (deferred to a later depth phase)\n\nKeep this playbook at scaffold depth. Explicitly **out of scope here**:\n\n- **Typed timeline / scheduling.** `targetWindow` is free prose and ordering is a\n coarse `sequence` integer. Start/end dates, date math, and calendar alignment\n are a later-phase concern — a phase says \"first 3 months post-launch\" as prose\n today, not a typed interval. Do NOT build a schedule model.\n- **Inter-phase dependency graph.** Phases are an ordered list, not a DAG. Edges\n between phases (phase B needs phase A's delivery) are a later-phase concern.\n- **Capacity / velocity / estimation and prioritization frameworks.** A phase is a\n themed bucket of deliver-refs, not an estimated, capacity-fitted, RICE/WSJF/MoSCoW-\n ranked plan. If you catch yourself estimating effort or building a prioritization\n model, **stop** — that is depth-phase work.\n- **A typed out-of-scope / explicitly-deferred axis** (see the scope decision\n above) — a `_core/roadmap` reshape, filed as a depth-phase item, not built here.\n- **Requirement priority + inter-requirement dependency.** A requirement has no\n MoSCoW/priority axis and no \"requirement A needs B\" edge; only the roadmap\n `sequence` approximates ordering. Do NOT reshape the requirements contract to add\n them — it is a filed depth-phase question.\n- **A dedicated non-functional-requirement delivery axis.** `deliversRequirementRefs`\n today carries functional AND non-functional refs against one combined pool. Some\n NFRs are cross-cutting quality attributes (accessibility, latency) that no single\n phase \"delivers\" — whether NFRs deserve their own phase-delivery axis (or a\n per-phase quality bar) rather than sharing `deliversRequirementRefs` is a\n depth-phase question. At scaffold depth, slot a phase-specific NFR into\n `deliversRequirementRefs` and leave universal NFRs unphased; do NOT reshape\n `_core` for a separate axis here.\n- **Per-phase ICP/segment targeting.** A phase cannot declare the primary\n ICP/segment it serves; that is a filed depth-phase gap. Note it in prose if it\n matters; do NOT reshape `_core` to add a typed edge.\n- **Deterministic feature → phase traceability.** Because a requirement carries no\n typed market pain/gain edge (and no typed feature → requirement edge), the\n feature → benefit → requirement → phase trace is prose-only. Whether the\n requirements family should gain a pain/gain edge is a filed depth-phase question —\n do NOT reshape the requirements or roadmap contract here.\n\nIf you find yourself perfecting any of the above, stop and leave the slice at\nscaffold depth — depth is a dedicated later pass done once, holistically.\n\n## What you emit\n\nOne artifact for this journey: the **roadmap** at\n`specifications/src/roadmap/index.ts` — a `RoadmapSnapshot` whose `phases`\nare the ordered, MVP-first releases, each with a distinct `sequence`, a coarse\n`horizon` + `kind`, and its `deliversRequirementRefs` / `deliversSuccessCriterionRefs`\nwired — passing the `roadmap-valid` gate.\n\nThe creator reviews the result (approval posture `approve_result`) before it is\naccepted as the journey's roadmap.\n\n## References\n\n- `.claude/skills/journey-status/SKILL.md` — read the current journey state and\n scope, and confirm a `market-snapshot` (the capability map) already exists to\n plan releases from.\n",
61
60
  "skillRefs": [
62
61
  "journey-status"
63
62
  ],
@@ -76,7 +75,9 @@
76
75
  ]
77
76
  },
78
77
  "contextSources": [],
79
- "suspendsForClarification": false
78
+ "suspendsForClarification": false,
79
+ "briefAdequacyGate": false,
80
+ "briefAdequacyExcludedSources": []
80
81
  },
81
82
  "executor": "companion_agent",
82
83
  "approval": "approve_result",