@atlashub/smartstack-cli 3.39.0 → 3.40.0

package/templates/skills/business-analyse/references/consolidation-structural-checks.md

@@ -1,107 +1,107 @@
-# Consolidation Structural Schema Conformity Checks
-
-> Post-generation validation. After semantic checks, verify ALL module feature.json files conform to JSON schema.
-
-## A. Required Sections Presence
-
-| Section | Required | Check |
-|---------|----------|-------|
-| `metadata` | YES | Must have: app, module, language, featureDescription, featureType |
-| `discovery` | YES | Must exist (even if minimal) |
-| `analysis.entities[]` | YES | Must have >=1 entity |
-| `analysis.businessRules[]` | YES | Must have >=1 BR |
-| `analysis.objectives[]` | YES | Must have >=1 objective |
-| `analysis.processFlow` | YES | Must have entryPoints, mainFlow, decisionPoints |
-| `analysis.dataLifecycle` | YES | Must have retention, archival, deletion, gdprRelevant |
-| `specification.useCases[]` | YES | Must have >=1 UC |
-| `specification.functionalRequirements[]` | YES | Must have >=1 FR |
-| `specification.permissionMatrix` | YES | Must be `{permissions[], roleAssignments[]}` (NOT flat array) |
-| `specification.sections[]` | YES | Must have >=1 section with resources |
-| `specification.uiWireframes[]` | YES | Must have >=1 wireframe with `layout` object |
-| `specification.lifeCycles[]` | IF entities have status | States must be objects (NOT strings) |
-| `specification.seedDataCore` | YES | Must have 7 arrays: navigationModules, navigationSections, navigationResources, navigationTranslations, permissions, rolePermissions, permissionConstants |
-| `specification.seedDataBusiness` | NO (WARNING) | Business seed data is generated during implementation (ralph-loop), not during BA. If present, validates format. |
-| `specification.gherkinScenarios` | YES | Must be `{feature, scenarios[]}` (NOT flat array) |
-| `specification.navigation` | YES | Must have entries[] |
-| `specification.i18nKeys[]` | YES | Must have >=1 key with fr, en, it, de |
-| `validation` | YES | Must have completenessChecks, consistencyChecks, semanticChecks, decision |
-
-## B. Field Name Conformity
-
-> **Two-phase entity schema:** In the **analysis** phase (step-03a2), entity attributes use a business-friendly format WITHOUT `type`.
-> In the **compilation** phase (step-03c), `type` is auto-inferred and added to each attribute.
-> Therefore `type` is FORBIDDEN in the raw analysis input but MANDATORY after compilation.
-
-| Section | FORBIDDEN fields | Correct fields |
-|---------|-----------------|----------------|
-| `entities[].attributes[]` | `values`, `rules` | `type` (added by step-03c compilation), `description`, `unique`, `validation` |
-| `businessRules[]` | `rule`, `condition` (singular) | `name`, `statement`, `conditions[]` |
-| `useCases[]` | `actor`, `linkedBRs`, `linkedFRs` | `primaryActor`, `linkedRules` |
-| `functionalRequirements[]` | `name`, `description`, `linkedUCs`, `linkedBRs` | `statement`, `linkedUseCases`, `linkedRules` |
-| `permissionMatrix` | flat array with `resource`/`roles` | `{permissions[], roleAssignments[]}` |
-| `lifeCycles[].states[]` | string values like `"Active"` | objects with `{id, displayName, color, isTerminal}` |
-| `seedDataCore` | flat object, string values | 7 typed arrays |
-| `gherkinScenarios` | flat scenario array | `{feature, scenarios[]}` nested |
-
-## C. ID Pattern Conformity
-
-| ID Type | Pattern | Example |
-|---------|---------|---------|
-| BR | `BR-(VAL\|CALC\|WF\|SEC\|DATA)-{PREFIX}-\d{3}` | BR-VAL-RM-001 |
-| UC | `UC-{PREFIX}-\d{3}` | UC-RM-001 |
-| FR | `FR-{PREFIX}-\d{3}` | FR-RM-001 |
-| Permissions (module) | `{app}.{module}.{action}` | freebike.repairs.create |
-| Permissions (section) | `{app}.{module}.{section}.{action}` | freebike.repairs.dashboard.read |
-
-## D. Cross-Module ID Uniqueness
-
-```
-SET all_ids = {}
-FOR each module:
-  FOR each BR, UC, FR id:
-    IF id IN all_ids -> ERROR: "Duplicate {id} in {module1} and {module2}"
-    ADD id to all_ids
-```
-
-## E. Wireframe Layout Completeness
-
-```
-IF missing componentMapping[] -> ERROR
-IF missing layout object -> ERROR
-IF missing layout.regions[] -> ERROR
-FOR each region: IF missing position or components[] -> ERROR
-```
-
-## F. SeedDataCore Translation Coverage
-
-```
-FOR each navigationModules[].code:
-  FOR each language in [fr, en, it, de]:
-    IF NOT EXISTS navigationTranslations[] with moduleCode AND language -> ERROR
-```
-
-## G. API Endpoint Coverage
-
-```
-FOR each entity in analysis.entities[]:
-  endpoints = specification.apiEndpoints[].filter(e => e.path.includes(entity.name.toLowerCase()))
-  IF endpoints.length < 5 -> WARNING: "Entity {entity.name} has <5 endpoints (expected: GET list, GET id, POST, PUT, DELETE)"
-  IF endpoints.length === 0 -> ERROR: "Entity {entity.name} has NO endpoints — missing controller"
-```
-
-## Result Aggregation
-
-```json
-{
-  "structuralChecks": [
-    { "check": "required-sections", "module": "...", "status": "PASS|ERROR", "details": "..." },
-    { "check": "field-names", "module": "...", "status": "PASS|ERROR", "details": "..." },
-    { "check": "id-patterns", "module": "...", "status": "PASS|ERROR", "details": "..." },
-    { "check": "id-uniqueness", "module": "...", "status": "PASS|ERROR", "details": "..." },
-    { "check": "wireframe-layout", "module": "...", "status": "PASS|ERROR", "details": "..." },
-    { "check": "seeddata-translations", "module": "...", "status": "PASS|ERROR", "details": "..." }
-  ]
-}
-```
-
-**Blocking rule:** 0 ERROR -> PASS. >=1 ERROR -> BLOCK (attempt auto-fix, re-write, re-check).
+# Consolidation Structural Schema Conformity Checks
+
+> Post-generation validation. After semantic checks, verify ALL module feature.json files conform to JSON schema.
+
+## A. Required Sections Presence
+
+| Section | Required | Check |
+|---------|----------|-------|
+| `metadata` | YES | Must have: app, module, language, featureDescription, featureType |
+| `discovery` | YES | Must exist (even if minimal) |
+| `analysis.entities[]` | YES | Must have >=1 entity |
+| `analysis.businessRules[]` | YES | Must have >=1 BR |
+| `analysis.objectives[]` | YES | Must have >=1 objective |
+| `analysis.processFlow` | YES | Must have entryPoints, mainFlow, decisionPoints |
+| `analysis.dataLifecycle` | YES | Must have retention, archival, deletion, gdprRelevant |
+| `specification.useCases[]` | YES | Must have >=1 UC |
+| `specification.functionalRequirements[]` | YES | Must have >=1 FR |
+| `specification.permissionMatrix` | YES | Must be `{permissions[], roleAssignments[]}` (NOT flat array) |
+| `specification.sections[]` | YES | Must have >=1 section with resources |
+| `specification.uiWireframes[]` | YES | Must have >=1 wireframe with `layout` object |
+| `specification.lifeCycles[]` | IF entities have status | States must be objects (NOT strings) |
+| `specification.seedDataCore` | YES | Must have 7 arrays: navigationModules, navigationSections, navigationResources, navigationTranslations, permissions, rolePermissions, permissionConstants |
+| `specification.seedDataBusiness` | NO (WARNING) | Business seed data is generated during implementation (ralph-loop), not during BA. If present, validates format. |
+| `specification.gherkinScenarios` | YES | Must be `{feature, scenarios[]}` (NOT flat array) |
+| `specification.navigation` | YES | Must have entries[] |
+| `specification.i18nKeys[]` | YES | Must have >=1 key with fr, en, it, de |
+| `validation` | YES | Must have completenessChecks, consistencyChecks, semanticChecks, decision |
+
+## B. Field Name Conformity
+
+> **Two-phase entity schema:** In the **analysis** phase (step-03a2), entity attributes use a business-friendly format WITHOUT `type`.
+> In the **compilation** phase (step-03c), `type` is auto-inferred and added to each attribute.
+> Therefore `type` is FORBIDDEN in the raw analysis input but MANDATORY after compilation.
+
+| Section | FORBIDDEN fields | Correct fields |
+|---------|-----------------|----------------|
+| `entities[].attributes[]` | `values`, `rules` | `type` (added by step-03c compilation), `description`, `unique`, `validation` |
+| `businessRules[]` | `rule`, `condition` (singular) | `name`, `statement`, `conditions[]` |
+| `useCases[]` | `actor`, `linkedBRs`, `linkedFRs` | `primaryActor`, `linkedRules` |
+| `functionalRequirements[]` | `name`, `description`, `linkedUCs`, `linkedBRs` | `statement`, `linkedUseCases`, `linkedRules` |
+| `permissionMatrix` | flat array with `resource`/`roles` | `{permissions[], roleAssignments[]}` |
+| `lifeCycles[].states[]` | string values like `"Active"` | objects with `{id, displayName, color, isTerminal}` |
+| `seedDataCore` | flat object, string values | 7 typed arrays |
+| `gherkinScenarios` | flat scenario array | `{feature, scenarios[]}` nested |
+
+## C. ID Pattern Conformity
+
+| ID Type | Pattern | Example |
+|---------|---------|---------|
+| BR | `BR-(VAL\|CALC\|WF\|SEC\|DATA)-{PREFIX}-\d{3}` | BR-VAL-RM-001 |
+| UC | `UC-{PREFIX}-\d{3}` | UC-RM-001 |
+| FR | `FR-{PREFIX}-\d{3}` | FR-RM-001 |
+| Permissions (module) | `{app}.{module}.{action}` | freebike.repairs.create |
+| Permissions (section) | `{app}.{module}.{section}.{action}` | freebike.repairs.dashboard.read |
+
+## D. Cross-Module ID Uniqueness
+
+```
+SET all_ids = {}
+FOR each module:
+  FOR each BR, UC, FR id:
+    IF id IN all_ids -> ERROR: "Duplicate {id} in {module1} and {module2}"
+    ADD id to all_ids
+```
+
+## E. Wireframe Layout Completeness
+
+```
+IF missing componentMapping[] -> ERROR
+IF missing layout object -> ERROR
+IF missing layout.regions[] -> ERROR
+FOR each region: IF missing position or components[] -> ERROR
+```
+
+## F. SeedDataCore Translation Coverage
+
+```
+FOR each navigationModules[].code:
+  FOR each language in [fr, en, it, de]:
+    IF NOT EXISTS navigationTranslations[] with moduleCode AND language -> ERROR
+```
+
+## G. API Endpoint Coverage
+
+```
+FOR each entity in analysis.entities[]:
+  endpoints = specification.apiEndpoints[].filter(e => e.path.includes(entity.name.toLowerCase()))
+  IF endpoints.length < 5 -> WARNING: "Entity {entity.name} has <5 endpoints (expected: GET list, GET id, POST, PUT, DELETE)"
+  IF endpoints.length === 0 -> ERROR: "Entity {entity.name} has NO endpoints — missing controller"
+```
+
+## Result Aggregation
+
+```json
+{
+  "structuralChecks": [
+    { "check": "required-sections", "module": "...", "status": "PASS|ERROR", "details": "..." },
+    { "check": "field-names", "module": "...", "status": "PASS|ERROR", "details": "..." },
+    { "check": "id-patterns", "module": "...", "status": "PASS|ERROR", "details": "..." },
+    { "check": "id-uniqueness", "module": "...", "status": "PASS|ERROR", "details": "..." },
+    { "check": "wireframe-layout", "module": "...", "status": "PASS|ERROR", "details": "..." },
+    { "check": "seeddata-translations", "module": "...", "status": "PASS|ERROR", "details": "..." }
+  ]
+}
+```
+
+**Blocking rule:** 0 ERROR -> PASS. >=1 ERROR -> BLOCK (attempt auto-fix, re-write, re-check).

package/templates/skills/business-analyse/references/deploy-data-build.md

@@ -1,180 +1,180 @@
-# FEATURE_DATA & EMBEDDED_ARTIFACTS Build Reference
-
-> Reference for step-05b-deploy.md — complete mapping specification for the interactive HTML.
-
-## FEATURE_DATA Object
-
-```javascript
-const FEATURE_DATA = {
-  metadata: {
-    applicationName: master.application_name,
-    applicationId: master.feature_id,
-    version: master.version,
-    createdAt: ISO_TIMESTAMP,
-    analysisMode: master.metadata.analysisMode || "interactive" // always "interactive" — analysis is always interactive
-  },
-  cadrage: {
-    // CONTEXT SECTION — merged from problem/asIs/toBe (lean cadrage)
-    context: {
-      problem: master.cadrage.problem || "",           // flat string from feature.json
-      trigger: master.cadrage.trigger || "",            // flat string from feature.json
-      currentSituation: master.cadrage.asIs || "",      // flat string from feature.json
-      desiredSituation: master.cadrage.toBe || "",      // flat string from feature.json
-      painPoints: (master.cadrage.stakeholders || []).flatMap(s => s.painPoints || []),
-      acceptanceCriteria: (master.cadrage.acceptanceCriteria || []).map(c => c.criterion)
-    },
-    scope: {
-      // CONVERT feature.json keys to HTML keys:
-      // mustHave → vital, shouldHave → important, couldHave → optional, outOfScope → excluded
-      // CRITICAL: Preserve per-item `module` assignment from coverageMatrix (NOT modules[0])
-      vital: (master.cadrage.globalScope?.mustHave || master.cadrage.coverageMatrix?.filter(i => i.category === "mustHave") || [])
-        .map(item => typeof item === 'string' ? { name: item, description: "", module: null } : { name: item.item || item, description: item.notes || "", module: item.module || null }),
-      important: (master.cadrage.globalScope?.shouldHave || master.cadrage.coverageMatrix?.filter(i => i.category === "shouldHave") || [])
-        .map(item => typeof item === 'string' ? { name: item, description: "", module: null } : { name: item.item || item, description: item.notes || "", module: item.module || null }),
-      optional: (master.cadrage.globalScope?.couldHave || master.cadrage.coverageMatrix?.filter(i => i.category === "couldHave") || [])
-        .map(item => typeof item === 'string' ? { name: item, description: "", module: null } : { name: item.item || item, description: item.notes || "", module: item.module || null }),
-      excluded: (master.cadrage.globalScope?.outOfScope || master.cadrage.coverageMatrix?.filter(i => i.category === "outOfScope") || [])
-        .map(item => typeof item === 'string' ? { name: item, description: "", module: null } : { name: item.item || item, description: item.notes || "", module: item.module || null })
-    },
-    risks: (master.cadrage.risks || []).map(r => ({
-      description: r.description,
-      probability: r.probability,
-      impact: r.impact,
-      mitigation: r.mitigation
-    })),
-    stakeholders: (master.cadrage.stakeholders || []).map(s => ({
-      role: s.role,
-      function: s.function,
-      tasks: s.tasks || [],
-      frequency: s.frequency || "",
-      access: s.involvement || "",
-      frustrations: (s.painPoints || []).join("\n")
-    }))
-  },
-  modules: [
-    // FOR EACH module in master.modules[]:
-    {
-      code: module.code,                              // PascalCase key (e.g., "GestionTemps")
-      name: module.name || module.code,               // Display name (e.g., "Gestion du temps") — MANDATORY
-      description: module.description || "",
-      featureType: module.featureType || "data-centric",
-      priority: module.priority || "must",
-      estimatedComplexity: module.estimatedComplexity || "medium",
-      entities: module.entities || [],
-      anticipatedSections: module.anticipatedSections || [],
-      dependencies: module.dependencies || [],
-      dependents: module.dependents || []
-    }
-  ],
-  moduleSpecs: {
-    // CRITICAL: Must have ONE entry per module with ALL module data
-    "{moduleCode}": {
-      useCases: module.specification.useCases || [],
-      businessRules: module.analysis.businessRules || [],
-      // ENTITY SAFETY NET: map fields[] → attributes[] if agent deviated from canonical format
-      entities: (module.analysis.entities || []).map(ent => ({
-        name: ent.name,
-        description: ent.description || "",
-        attributes: (ent.attributes || []).length > 0
-          ? ent.attributes.map(a => ({ name: a.name, description: a.description || "" }))
-          : (ent.fields || []).map(f => ({ name: f.name, description: f.description || f.type || "" })),
-        relationships: (ent.relationships || []).length > 0
-          ? ent.relationships.map(r =>
-              typeof r === 'string' ? r : `${r.target} (${r.type}) - ${r.description || ""}`)
-          : (ent.fields || []).filter(f => f.foreignKey)
-              .map(f => `${f.foreignKey.table} (N:1) - ${f.description || f.name}`)
-      })),
-      permissions: module.specification.permissions || [],
-      apiEndpoints: module.specification.apiEndpoints || []
-    }
-  },
-  consolidation: {
-    integrations: master.consolidation.integrations || [],
-    sharedEntities: master.consolidation.sharedEntities || [],
-    sequenceDiagrams: master.consolidation.e2eFlows || []
-  },
-  handoff: {
-    complexity: master.handoff.complexity,
-    implementationStrategy: master.handoff.implementationStrategy,
-    moduleOrder: master.handoff.moduleOrder,
-    filesToCreate: master.handoff.filesToCreate,
-    brToCodeMapping: master.handoff.brToCodeMapping,
-    apiEndpointSummary: master.handoff.apiEndpointSummary
-  }
-};
-```
-
-### Build Process
-
-1. Extract metadata from master feature.json
-2. Extract cadrage from master feature.json (CONVERT scope keys)
-3. Extract stakeholders from master.stakeholders
-4. Iterate ALL modules and populate moduleSpecs (THIS IS CRITICAL — empty moduleSpecs = BUG)
-5. For EACH module, extract: useCases, businessRules, entities, permissions, apiEndpoints
-6. Extract consolidation data (integrations, shared entities, E2E flows)
-7. Extract handoff section (complexity, strategy, module order, file counts)
-
-## EMBEDDED_ARTIFACTS Object
-
-> **CRITICAL:** The wireframes object is keyed by moduleCode (NOT a flat array).
-> Field names MUST be renamed from feature.json: `mockupFormat` → `format`, `mockup` → `content`.
-> The HTML renderer reads `wf.format` and `wf.content` — using feature.json names will produce EMPTY mockups.
-
-```javascript
-const EMBEDDED_ARTIFACTS = {
-  wireframes: {
-    // PER-MODULE keyed object (NOT a flat array)
-    // FOR EACH module: extract from specification.uiWireframes[] OR specification.wireframes[] (SAFETY NET)
-    // IMPORTANT: The agent may write wireframes under either key name — always check BOTH
-    [moduleCode]: (moduleFeature.specification.uiWireframes || moduleFeature.specification.wireframes || []).map(wf => ({
-      screen: wf.screen || wf.name || wf.title || wf.id || "",  // SAFETY NET: fallback name/title/id → screen
-      section: wf.section || "",                      // e.g. "list"
-      format: wf.mockupFormat || "ascii",             // RENAME: mockupFormat → format
-      content: wf.mockup || wf.ascii || wf.content || "",  // SAFETY NET: mockup/ascii/content → content
-      svgContent: null,                              // Populated by SVG generation step (see wireframe-svg-style-guide.md)
-      description: wf.description || "",
-      elements: wf.elements || [],                    // [{ id, type, label }] or ["DataGrid", ...]
-      actions: wf.actions || [],                      // [{ trigger, action }] or ["filter", ...]
-      // SAFETY NET: convert object → array format if agent used { "Header": "PageHeader" } instead of [{ wireframeElement, reactComponent }]
-      componentMapping: Array.isArray(wf.componentMapping) ? wf.componentMapping
-        : typeof wf.componentMapping === 'object' && wf.componentMapping !== null
-          ? Object.entries(wf.componentMapping).map(([k, v]) => ({ wireframeElement: k, reactComponent: v }))
-          : [],
-      layout: typeof wf.layout === 'object' ? wf.layout : null,  // SAFETY NET: ignore string layout ("grid")
-      permissionsRequired: wf.permissionsRequired || []
-    }))
-  },
-  e2eFlows: (master.consolidation?.e2eFlows || []).map(flow => ({
-    name: flow.name,
-    diagram: flow.steps.map(s => `${s.action}(${s.module})`).join(" ──→ "),
-    steps: flow.steps || [],
-    actors: [...new Set(flow.steps.map(s => s.permission?.split(".")[0]).filter(Boolean))].join(", "),
-    modules: [...new Set(flow.steps.map(s => s.module))].join(" → ")
-  })),
-  dependencyGraph: {
-    nodes: (master.modules || []).map(m => ({
-      id: m.code, label: m.code, type: m.featureType || "data-centric"
-    })),
-    edges: (master.dependencyGraph?.edges || []).map(e => ({
-      from: e.from, to: e.to, description: e.description || ""
-    }))
-  }
-};
-```
-
-### Artifact Gathering
-
-1. For EACH module: read `specification.uiWireframes[]` OR `specification.wireframes[]` (check BOTH keys — agent may use either), **rename fields** (`mockupFormat`→`format`, `mockup`/`ascii`/`content`→`content`, `screen`/`name`/`title`→`screen`), store under `wireframes[moduleCode]`
-2. Read master's `consolidation.e2eFlows[]` and build e2eFlows array with diagram generation
-3. Read master's `dependencyGraph` and build nodes/edges
-4. Serialize as JSON with 2-space indentation
-5. **SVG Wireframe Enrichment** (after wireframe extraction): For EACH wireframe with `content` (ASCII) and `svgContent === null`, generate SVG using parallel Task(sonnet) agents. See [wireframe-svg-style-guide.md](wireframe-svg-style-guide.md) for the complete prompt template and orchestration process. If generation fails for any wireframe → leave `svgContent` as null (graceful fallback to ASCII).
-
-## Placeholder Replacement
-
-1. Serialize FEATURE_DATA as JSON (2-space indent)
-2. Serialize EMBEDDED_ARTIFACTS as JSON (2-space indent)
-3. Replace `{{FEATURE_DATA}}` with serialized FEATURE_DATA
-4. Replace `{{EMBEDDED_ARTIFACTS}}` with serialized EMBEDDED_ARTIFACTS
-5. Replace `{{APPLICATION_NAME}}`, `{{APPLICATION_ID}}`, `{{VERSION}}`, `{{CREATED_AT}}`
-6. Verify: grep for `{{` to confirm all placeholders replaced
+# FEATURE_DATA & EMBEDDED_ARTIFACTS Build Reference
+
+> Reference for step-05b-deploy.md — complete mapping specification for the interactive HTML.
+
+## FEATURE_DATA Object
+
+```javascript
+const FEATURE_DATA = {
+  metadata: {
+    applicationName: master.application_name,
+    applicationId: master.feature_id,
+    version: master.version,
+    createdAt: ISO_TIMESTAMP,
+    analysisMode: master.metadata.analysisMode || "interactive" // always "interactive" — analysis is always interactive
+  },
+  cadrage: {
+    // CONTEXT SECTION — merged from problem/asIs/toBe (lean cadrage)
+    context: {
+      problem: master.cadrage.problem || "",           // flat string from feature.json
+      trigger: master.cadrage.trigger || "",            // flat string from feature.json
+      currentSituation: master.cadrage.asIs || "",      // flat string from feature.json
+      desiredSituation: master.cadrage.toBe || "",      // flat string from feature.json
+      painPoints: (master.cadrage.stakeholders || []).flatMap(s => s.painPoints || []),
+      acceptanceCriteria: (master.cadrage.acceptanceCriteria || []).map(c => c.criterion)
+    },
+    scope: {
+      // CONVERT feature.json keys to HTML keys:
+      // mustHave → vital, shouldHave → important, couldHave → optional, outOfScope → excluded
+      // CRITICAL: Preserve per-item `module` assignment from coverageMatrix (NOT modules[0])
+      vital: (master.cadrage.globalScope?.mustHave || master.cadrage.coverageMatrix?.filter(i => i.category === "mustHave") || [])
+        .map(item => typeof item === 'string' ? { name: item, description: "", module: null } : { name: item.item || item, description: item.notes || "", module: item.module || null }),
+      important: (master.cadrage.globalScope?.shouldHave || master.cadrage.coverageMatrix?.filter(i => i.category === "shouldHave") || [])
+        .map(item => typeof item === 'string' ? { name: item, description: "", module: null } : { name: item.item || item, description: item.notes || "", module: item.module || null }),
+      optional: (master.cadrage.globalScope?.couldHave || master.cadrage.coverageMatrix?.filter(i => i.category === "couldHave") || [])
+        .map(item => typeof item === 'string' ? { name: item, description: "", module: null } : { name: item.item || item, description: item.notes || "", module: item.module || null }),
+      excluded: (master.cadrage.globalScope?.outOfScope || master.cadrage.coverageMatrix?.filter(i => i.category === "outOfScope") || [])
+        .map(item => typeof item === 'string' ? { name: item, description: "", module: null } : { name: item.item || item, description: item.notes || "", module: item.module || null })
+    },
+    risks: (master.cadrage.risks || []).map(r => ({
+      description: r.description,
+      probability: r.probability,
+      impact: r.impact,
+      mitigation: r.mitigation
+    })),
+    stakeholders: (master.cadrage.stakeholders || []).map(s => ({
+      role: s.role,
+      function: s.function,
+      tasks: s.tasks || [],
+      frequency: s.frequency || "",
+      access: s.involvement || "",
+      frustrations: (s.painPoints || []).join("\n")
+    }))
+  },
+  modules: [
+    // FOR EACH module in master.modules[]:
+    {
+      code: module.code,                              // PascalCase key (e.g., "GestionTemps")
+      name: module.name || module.code,               // Display name (e.g., "Gestion du temps") — MANDATORY
+      description: module.description || "",
+      featureType: module.featureType || "data-centric",
+      priority: module.priority || "must",
+      estimatedComplexity: module.estimatedComplexity || "medium",
+      entities: module.entities || [],
+      anticipatedSections: module.anticipatedSections || [],
+      dependencies: module.dependencies || [],
+      dependents: module.dependents || []
+    }
+  ],
+  moduleSpecs: {
+    // CRITICAL: Must have ONE entry per module with ALL module data
+    "{moduleCode}": {
+      useCases: module.specification.useCases || [],
+      businessRules: module.analysis.businessRules || [],
+      // ENTITY SAFETY NET: map fields[] → attributes[] if agent deviated from canonical format
+      entities: (module.analysis.entities || []).map(ent => ({
+        name: ent.name,
+        description: ent.description || "",
+        attributes: (ent.attributes || []).length > 0
+          ? ent.attributes.map(a => ({ name: a.name, description: a.description || "" }))
+          : (ent.fields || []).map(f => ({ name: f.name, description: f.description || f.type || "" })),
+        relationships: (ent.relationships || []).length > 0
+          ? ent.relationships.map(r =>
+              typeof r === 'string' ? r : `${r.target} (${r.type}) - ${r.description || ""}`)
+          : (ent.fields || []).filter(f => f.foreignKey)
+              .map(f => `${f.foreignKey.table} (N:1) - ${f.description || f.name}`)
+      })),
+      permissions: module.specification.permissions || [],
+      apiEndpoints: module.specification.apiEndpoints || []
+    }
+  },
+  consolidation: {
+    integrations: master.consolidation.integrations || [],
+    sharedEntities: master.consolidation.sharedEntities || [],
+    sequenceDiagrams: master.consolidation.e2eFlows || []
+  },
+  handoff: {
+    complexity: master.handoff.complexity,
+    implementationStrategy: master.handoff.implementationStrategy,
+    moduleOrder: master.handoff.moduleOrder,
+    filesToCreate: master.handoff.filesToCreate,
+    brToCodeMapping: master.handoff.brToCodeMapping,
+    apiEndpointSummary: master.handoff.apiEndpointSummary
+  }
+};
+```
+
+### Build Process
+
+1. Extract metadata from master feature.json
+2. Extract cadrage from master feature.json (CONVERT scope keys)
+3. Extract stakeholders from master.stakeholders
+4. Iterate ALL modules and populate moduleSpecs (THIS IS CRITICAL — empty moduleSpecs = BUG)
+5. For EACH module, extract: useCases, businessRules, entities, permissions, apiEndpoints
+6. Extract consolidation data (integrations, shared entities, E2E flows)
+7. Extract handoff section (complexity, strategy, module order, file counts)
+
+## EMBEDDED_ARTIFACTS Object
+
+> **CRITICAL:** The wireframes object is keyed by moduleCode (NOT a flat array).
+> Field names MUST be renamed from feature.json: `mockupFormat` → `format`, `mockup` → `content`.
+> The HTML renderer reads `wf.format` and `wf.content` — using feature.json names will produce EMPTY mockups.
+
+```javascript
+const EMBEDDED_ARTIFACTS = {
+  wireframes: {
+    // PER-MODULE keyed object (NOT a flat array)
+    // FOR EACH module: extract from specification.uiWireframes[] OR specification.wireframes[] (SAFETY NET)
+    // IMPORTANT: The agent may write wireframes under either key name — always check BOTH
+    [moduleCode]: (moduleFeature.specification.uiWireframes || moduleFeature.specification.wireframes || []).map(wf => ({
+      screen: wf.screen || wf.name || wf.title || wf.id || "",  // SAFETY NET: fallback name/title/id → screen
+      section: wf.section || "",                      // e.g. "list"
+      format: wf.mockupFormat || "ascii",             // RENAME: mockupFormat → format
+      content: wf.mockup || wf.ascii || wf.content || "",  // SAFETY NET: mockup/ascii/content → content
+      svgContent: null,                              // Populated by SVG generation step (see wireframe-svg-style-guide.md)
+      description: wf.description || "",
+      elements: wf.elements || [],                    // [{ id, type, label }] or ["DataGrid", ...]
+      actions: wf.actions || [],                      // [{ trigger, action }] or ["filter", ...]
+      // SAFETY NET: convert object → array format if agent used { "Header": "PageHeader" } instead of [{ wireframeElement, reactComponent }]
+      componentMapping: Array.isArray(wf.componentMapping) ? wf.componentMapping
+        : typeof wf.componentMapping === 'object' && wf.componentMapping !== null
+          ? Object.entries(wf.componentMapping).map(([k, v]) => ({ wireframeElement: k, reactComponent: v }))
+          : [],
+      layout: typeof wf.layout === 'object' ? wf.layout : null,  // SAFETY NET: ignore string layout ("grid")
+      permissionsRequired: wf.permissionsRequired || []
+    }))
+  },
+  e2eFlows: (master.consolidation?.e2eFlows || []).map(flow => ({
+    name: flow.name,
+    diagram: flow.steps.map(s => `${s.action}(${s.module})`).join(" ──→ "),
+    steps: flow.steps || [],
+    actors: [...new Set(flow.steps.map(s => s.permission?.split(".")[0]).filter(Boolean))].join(", "),
+    modules: [...new Set(flow.steps.map(s => s.module))].join(" → ")
+  })),
+  dependencyGraph: {
+    nodes: (master.modules || []).map(m => ({
+      id: m.code, label: m.code, type: m.featureType || "data-centric"
+    })),
+    edges: (master.dependencyGraph?.edges || []).map(e => ({
+      from: e.from, to: e.to, description: e.description || ""
+    }))
+  }
+};
+```
+
+### Artifact Gathering
+
+1. For EACH module: read `specification.uiWireframes[]` OR `specification.wireframes[]` (check BOTH keys — agent may use either), **rename fields** (`mockupFormat`→`format`, `mockup`/`ascii`/`content`→`content`, `screen`/`name`/`title`→`screen`), store under `wireframes[moduleCode]`
+2. Read master's `consolidation.e2eFlows[]` and build e2eFlows array with diagram generation
+3. Read master's `dependencyGraph` and build nodes/edges
+4. Serialize as JSON with 2-space indentation
+5. **SVG Wireframe Enrichment** (after wireframe extraction): For EACH wireframe with `content` (ASCII) and `svgContent === null`, generate SVG using parallel Task(sonnet) agents. See [wireframe-svg-style-guide.md](wireframe-svg-style-guide.md) for the complete prompt template and orchestration process. If generation fails for any wireframe → leave `svgContent` as null (graceful fallback to ASCII).
+
+## Placeholder Replacement
+
+1. Serialize FEATURE_DATA as JSON (2-space indent)
+2. Serialize EMBEDDED_ARTIFACTS as JSON (2-space indent)
+3. Replace `{{FEATURE_DATA}}` with serialized FEATURE_DATA
+4. Replace `{{EMBEDDED_ARTIFACTS}}` with serialized EMBEDDED_ARTIFACTS
+5. Replace `{{APPLICATION_NAME}}`, `{{APPLICATION_ID}}`, `{{VERSION}}`, `{{CREATED_AT}}`
+6. Verify: grep for `{{` to confirm all placeholders replaced