# Action Plan Workflow Hardening Implementation Plan > **READ FIRST — [`2026-08-06-plan-review-corrections.md`](../specs/2026-08-06-plan-review-corrections.md) is a BINDING addendum to this document.** Five independent reviewers found defects that were then verified against the repository. Where the addendum conflicts with anything below, the addendum wins. > **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development (recommended) or superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (`- [ ]`) syntax for tracking. **Goal:** Make every AI-generated follow-up step independently reviewable, assignable, prioritised, actionable and auditable while preserving the existing safe approval/checklist workflow. **Architecture:** Keep the 1:1 `ZoomMeetingActionPlan` and its JSON review snapshot, but expand each item with validated metadata and per-step assignee IDs. Approval still materialises normal `LeadActionItem` rows in one locked transaction. AI revisions are a separate proposal-only endpoint; Dashboard team data is lazy-loaded and role-scoped. **Tech Stack:** Laravel/PHP, Eloquent/MySQL, Vue 3/Inertia, Vitest, PHPUnit, existing `AiClient`, existing Lead visibility and Action Item repositories. ## Global Constraints - Read `docs/superpowers/specs/2026-08-06-ai-sales-follow-up-workspace-design.md` completely before editing. - Work on an isolated task branch/worktree based on the current local `dev-chen`; integrate verified task commits back into local `dev-chen` only. - Do not push and do not open a PR. - `features.zoom_action_plan_demo` remains false by default and gates every new endpoint/prop/control. - No due-date column. Rename user-facing “Today's Checklist” to “My Checklist”; Current Tasks means every open assigned item. - No automatic approval, assignment, customer contact, prompt tuning or persisted AI revision. - Add no request-rate limit in this scope. - Configure new prompt keys locally through the existing AI Prompt pinning mechanism to OpenAI `gpt-5.6-terra`; never hard-code provider/model in request controllers and never commit the local prompt-pin row. - Database enums use unsigned integers plus model constants/metadata arrays. - Preserve `meeting_report.next_steps` as strings; add `recommended_actions` without breaking Call/F2F/Zoom consumers. - TDD each task; run its focused PHP and JS tests before its commit. - Each task receives a fresh code review; database/security tasks also receive database/security review, and Vue tasks receive TypeScript/JavaScript review. - Do not commit production customer data, transcripts, prompt responses, credentials, `.env`, screenshots or browser artifacts. --- ## File map **Domain and schema** - Create `database/migrations/2026_08_06_100001_add_workflow_metadata_to_lead_action_items.php` — nullable generated-task priority/type fields. - Create `database/migrations/2026_08_06_100002_add_snapshots_to_lead_action_item_feedback.php` — rating-time evidence snapshots. - Modify `src/Lead/LeadActionItem.php` — integer priority/type constants, casts and serialized fields. - Modify `src/Lead/LeadActionItemFeedback.php` — snapshot fillable/casts. - Create `src/Conversation/Support/ConversationAnalysisHash.php` — deterministic canonical analysis hash used by feedback provenance. - Modify `src/Conversation/ConversationAnalysis.php` — additive recommended-action normalization. - Modify `src/Conversation/Concerns/HasConversationPerformance.php` — normalized recommended actions with legacy fallback. - Modify `resources/prompts/conversation_analysis.md` — additive JSON contract and evidence rules. **Action Plan backend** - Modify `src/Zoom/Services/ZoomActionPlanDraftService.php` — draft rows with AI metadata and default per-step assignee. - Modify `src/Zoom/Repositories/ZoomMeetingActionPlanRepository.php` — format/validate per-step metadata and materialise it. - Modify `app/Http/Requests/Manage/Zoom/ActionPlans/UpdateRequest.php` — per-row validation. - Modify `app/Http/Controllers/Manage/Zoom/ZoomActionPlansController.php` — UUID boundary mapping. - Create `app/Http/Requests/Manage/Zoom/ActionPlans/SuggestionRequest.php` — bounded revision request. - Create `app/Http/Controllers/Manage/Zoom/ActionPlanSuggestionController.php` — proposal-only AI endpoint. - Create `src/Zoom/Services/ActionPlanSuggestionContextBuilder.php` — current unsaved rows plus meeting snapshot. - Create `resources/prompts/action_plan_revision.md` and register `PROMPT_ACTION_PLAN_REVISION`. - Modify `routes/web.php` — one flag/auth-gated suggestion route. **Action Plan frontend** - Modify `resources/js/composables/useActionPlanEditor.js` and test — full row state/default apply/proposal apply. - Modify `resources/js/Components/Zoom/ActionPlanReviewModal.vue` — per-row controls and revision preview. - Create `resources/js/Components/Zoom/ActionPlanRevisionPreview.vue` — current/proposed diff with explicit Apply. **Feedback** - Modify `src/Lead/Repositories/LeadActionItemRepository.php` — snapshot upsert. - Modify `app/Http/Controllers/Manage/Leads/ActionItemsController.php` — map snapshot values server-side. - Modify `app/Http/Controllers/Manage/Zoom/ActionPlanFeedbackController.php` and `resources/js/Pages/Manage/Zoom/ActionPlans/Feedback.vue` — show rated snapshot/source metadata. **Checklist and contact actions** - Modify `app/Http/Controllers/Manage/DashboardController.php` — priority payload, lazy Super Admin team endpoint, stranded count. - Create `app/Http/Requests/Manage/Dashboard/TeamChecklistRequest.php` — filters/cursor validation. - Modify `resources/js/Pages/Manage/Dashboard.vue` and `resources/js/Components/TodaysChecklistModal.vue` (keep filename; rename visible copy) — My/Team workspace. - Create `resources/js/composables/useLeadContactActions.js` and test — shared safe call/WhatsApp handoff. - Modify `resources/js/Components/Sales/LeadCell.vue` — consume the shared contact composable. - Modify `routes/web.php` — team checklist route. --- ### Task 1: Add model-backed priority, type and feedback snapshot columns **Files:** - Create: `database/migrations/2026_08_06_100001_add_workflow_metadata_to_lead_action_items.php` - Create: `database/migrations/2026_08_06_100002_add_snapshots_to_lead_action_item_feedback.php` - Modify: `src/Lead/LeadActionItem.php` - Modify: `src/Lead/LeadActionItemFeedback.php` - Test: `tests/Feature/Zoom/ZoomActionPlanFoundationTest.php` - Test: `tests/Feature/Manage/Leads/ActionItemFeedbackTest.php` **Interfaces:** - Produces: `LeadActionItem::PRIORITY_HIGH|MEDIUM|LOW`, `PRIORITIES`, `TYPE_*`, `TYPES`. - Produces nullable integer columns `suggested_priority`, `priority`, `action_type`. - Produces feedback snapshots `body_snapshot`, `suggested_priority_snapshot`, `priority_snapshot`, `action_type_snapshot`, `analysis_hash_snapshot`. - [ ] **Step 1: Write failing schema/model tests** Add assertions that the new columns exist, constants are unsigned-integer-compatible, casts return integers, and legacy rows accept null metadata. Add a feedback test proving body/priority/type/hash snapshots survive source-item edits and soft deletion. ```php $item = $this->createGeneratedActionItem([ 'suggested_priority' => LeadActionItem::PRIORITY_HIGH, 'priority' => LeadActionItem::PRIORITY_MEDIUM, 'action_type' => LeadActionItem::TYPE_WHATSAPP, ]); $this->assertSame(LeadActionItem::PRIORITY_MEDIUM, $item->priority); $this->assertSame('Medium', LeadActionItem::PRIORITIES[$item->priority]['name']); ``` - [ ] **Step 2: Run focused tests and verify failure** Run: ```bash php artisan test tests/Feature/Zoom/ZoomActionPlanFoundationTest.php tests/Feature/Manage/Leads/ActionItemFeedbackTest.php ``` Expected: FAIL because constants/columns do not exist. - [ ] **Step 3: Add the migrations and constants** Use nullable columns so old/manual Action Items are unchanged: ```php $table->unsignedInteger('suggested_priority')->nullable()->index()->after('status'); $table->unsignedInteger('priority')->nullable()->index()->after('suggested_priority'); $table->unsignedInteger('action_type')->nullable()->index()->after('priority'); ``` Use constants: ```php public const PRIORITY_HIGH = 1; public const PRIORITY_MEDIUM = 2; public const PRIORITY_LOW = 3; public const TYPE_CALL = 1; public const TYPE_WHATSAPP = 2; public const TYPE_SEND_INFORMATION = 3; public const TYPE_SCHEDULE = 4; public const TYPE_INTERNAL = 5; public const TYPE_OTHER = 6; ``` The migration `down()` drops only the columns created by that migration. Add integer casts and fillable fields explicitly. - [ ] **Step 4: Run migration round-trip and tests** Run the project scratch-database migration procedure, then the focused tests. Expected: migrations up/down succeed and tests PASS. - [ ] **Step 5: Review and commit** Require database review for reversible migrations, indexes, null behavior and integer constants. Fix findings, then commit: ```bash git add database/migrations/2026_08_06_100001_add_workflow_metadata_to_lead_action_items.php database/migrations/2026_08_06_100002_add_snapshots_to_lead_action_item_feedback.php src/Lead/LeadActionItem.php src/Lead/LeadActionItemFeedback.php tests/Feature/Zoom/ZoomActionPlanFoundationTest.php tests/Feature/Manage/Leads/ActionItemFeedbackTest.php git commit -m "feat(leads): add action item workflow metadata" ``` ### Task 2: Add backward-compatible AI recommended actions **Files:** - Modify: `resources/prompts/conversation_analysis.md` - Modify: `src/Conversation/ConversationAnalysis.php` - Modify: `src/Conversation/Concerns/HasConversationPerformance.php` - Test: `tests/Feature/Conversation/ConversationAnalyzerTest.php` - Test: `tests/Feature/Zoom/ZoomActionPlanDraftTest.php` **Interfaces:** - Produces: `ConversationAnalysis::PRIORITIES`, `ACTION_TYPES`. - Produces: `HasConversationPerformance::recommendedActionItems(): array`. - Preserves: `actionItems(): array` and every existing `next_steps` consumer. - [ ] **Step 1: Write failing normalizer and fallback tests** Cover valid rows, invalid enum clamping, nested/scalar garbage, maximum list length, missing field and a legacy analysis with only `next_steps`. ```php $analysis = ConversationAnalysis::normalize([ 'meeting_report' => [ 'next_steps' => ['Send comparison'], 'recommended_actions' => [[ 'body' => 'Send comparison', 'priority' => 'urgent', 'action_type' => 'magic', 'reason' => 'Financing concern', ]], ], ]); $this->assertSame('medium', $analysis['meeting_report']['recommended_actions'][0]['priority']); $this->assertSame('other', $analysis['meeting_report']['recommended_actions'][0]['action_type']); ``` - [ ] **Step 2: Run focused tests and verify failure** ```bash php artisan test tests/Feature/Conversation/ConversationAnalyzerTest.php tests/Feature/Zoom/ZoomActionPlanDraftTest.php ``` Expected: FAIL because recommended actions are currently discarded inside the core `meeting_report` shape. - [ ] **Step 3: Implement the additive prompt contract and normalizer** Keep `next_steps` unchanged and add: ```json "recommended_actions": [{ "body": "a concrete assignable action", "priority": "high | medium | low", "action_type": "call | whatsapp | send_information | schedule | internal | other", "reason": "one short evidence-based reason" }] ``` Prompt rules must say priority is urgency/importance, not conversion probability; reason must cite recorded needs/concerns/commitments; unknown evidence means omit the action rather than invent it. Normalize at most 10 non-empty rows. Implement a private `recommendedActions()` sanitizer and public trait method. When the array is empty, the trait maps legacy `next_steps` to Medium/Other/null-reason rows without mutating stored analysis. - [ ] **Step 4: Run regression tests for all three channels** ```bash php artisan test tests/Feature/Conversation/ConversationAnalyzerTest.php tests/Feature/Zoom/ZoomActionPlanDraftTest.php tests/Feature/Manage/Calls tests/Feature/Manage/F2f ``` Expected: PASS; existing string action items remain unchanged. - [ ] **Step 5: Review and commit** Require AI-output normalization and backward-compatibility review. Commit: ```bash git add resources/prompts/conversation_analysis.md src/Conversation/ConversationAnalysis.php src/Conversation/Concerns/HasConversationPerformance.php tests/Feature/Conversation/ConversationAnalyzerTest.php tests/Feature/Zoom/ZoomActionPlanDraftTest.php git commit -m "feat(ai): suggest typed action plan steps" ``` ### Task 3: Build and save per-step review metadata **Files:** - Modify: `src/Zoom/Services/ZoomActionPlanDraftService.php` - Modify: `src/Zoom/Repositories/ZoomMeetingActionPlanRepository.php` - Modify: `app/Http/Requests/Manage/Zoom/ActionPlans/UpdateRequest.php` - Modify: `app/Http/Controllers/Manage/Zoom/ZoomActionPlansController.php` - Test: `tests/Feature/Zoom/ZoomActionPlanDraftTest.php` - Test: `tests/Feature/Manage/Zoom/ZoomActionPlansEndpointsTest.php` **Interfaces:** - HTTP item: `{key,body,reason,suggested_priority,priority,action_type,assignee_uuid}`. - Stored item: same semantic values, with integer `assignee_id` and integer enum values. - Existing plan `assignee_id` remains the default-assignee field. - [ ] **Step 1: Write failing draft and endpoint tests** Cover AI metadata, legacy defaults, one assignee per row, plan-level default, forged/ineligible assignee, invalid priority/type, reanalysis replacement, and approved snapshot immutability. - [ ] **Step 2: Verify red tests** ```bash php artisan test tests/Feature/Zoom/ZoomActionPlanDraftTest.php tests/Feature/Manage/Zoom/ZoomActionPlansEndpointsTest.php ``` Expected: FAIL on missing row metadata and request rules. - [ ] **Step 3: Update draft construction and server formatting** Map codes with model constants and build rows: ```php $items[] = [ 'key' => $key, 'body' => $action['body'], 'reason' => $action['reason'], 'suggested_priority' => LeadActionItem::priorityFromCode($action['priority']), 'priority' => LeadActionItem::priorityFromCode($action['priority']), 'action_type' => LeadActionItem::typeFromCode($action['action_type']), 'assignee_id' => $defaultAssigneeId, ]; ``` Do not write on GET. Upgrade legacy rows only in serializer memory and on the next explicit PUT. - [ ] **Step 4: Validate/map every HTTP field explicitly** Add per-item rules, but resolve eligible users and enum constants again in the repository. `show()` maps stored `assignee_id` to UUID using an eager-loaded eligible-user map. `mapPlanInput()` maps fields one by one; never pass `validated()` wholesale. - [ ] **Step 5: Run tests, review and commit** ```bash php artisan test tests/Feature/Zoom/ZoomActionPlanDraftTest.php tests/Feature/Manage/Zoom/ZoomActionPlansEndpointsTest.php ``` Require security review of UUID/Lead visibility checks, then commit: ```bash git add src/Zoom/Services/ZoomActionPlanDraftService.php src/Zoom/Repositories/ZoomMeetingActionPlanRepository.php app/Http/Requests/Manage/Zoom/ActionPlans/UpdateRequest.php app/Http/Controllers/Manage/Zoom/ZoomActionPlansController.php tests/Feature/Zoom/ZoomActionPlanDraftTest.php tests/Feature/Manage/Zoom/ZoomActionPlansEndpointsTest.php git commit -m "feat(zoom): review action plan steps independently" ``` ### Task 4: Approve independently assigned and prioritised steps **Files:** - Modify: `src/Zoom/Repositories/ZoomMeetingActionPlanRepository.php` - Modify: `src/Lead/Repositories/LeadActionItemRepository.php` - Test: `tests/Feature/Zoom/ZoomActionPlanApprovalTest.php` **Interfaces:** - Consumes the stored per-step contract from Task 3. - Produces one `LeadActionItem` per row with its own pivot assignee and metadata. - [ ] **Step 1: Write failing approval tests** Cover mixed assignees, mixed priorities/types, one invalid row rolling back all rows, missing row assignee, double approval, simulated concurrency, unchanged notifier behavior and legacy draft approval. - [ ] **Step 2: Verify failures** ```bash php artisan test tests/Feature/Zoom/ZoomActionPlanApprovalTest.php ``` - [ ] **Step 3: Implement transaction-wide validation before writes** Resolve all row assignees and enum values into an in-memory `$resolvedSteps` array before calling `createGeneratedItems()`. Only after every row is valid may the repository write: ```php $data['lead_action_item'] = [ 'lead_id' => $lead->id, 'body' => $step['body'], 'suggested_priority' => $step['suggested_priority'], 'priority' => $step['priority'], 'action_type' => $step['action_type'], 'source_action_plan_id' => $plan->id, 'source_step_key' => $step['key'], ]; $data['assignee_ids'] = [$step['assignee']->id]; ``` Retain the row lock, transaction and unique index. - [ ] **Step 4: Run tests and review transaction safety** Expected: all approval tests PASS. Require database/security review for partial-write, forged UUID and idempotency paths. - [ ] **Step 5: Commit** ```bash git add src/Zoom/Repositories/ZoomMeetingActionPlanRepository.php src/Lead/Repositories/LeadActionItemRepository.php tests/Feature/Zoom/ZoomActionPlanApprovalTest.php git commit -m "feat(zoom): assign and prioritise approved steps" ``` ### Task 5: Upgrade the review editor UI **Files:** - Modify: `resources/js/composables/useActionPlanEditor.js` - Modify: `resources/js/composables/useActionPlanEditor.test.js` - Modify: `resources/js/Components/Zoom/ActionPlanReviewModal.vue` - Test: existing Vitest suite plus focused component coverage where the project pattern supports it. **Interfaces:** - Editor exposes `defaultAssigneeUuid`, `applyDefaultAssignee()`, `updateStep(index, patch)` and `replaceWithProposal(items)`. - `payload()` exactly matches Task 3's HTTP item contract. - [ ] **Step 1: Write failing composable tests** Prove init/defaults, row override, Apply to all, reorder identity, proposal replacement, invalid/empty rows and payload trimming. “Apply to all” is explicit; changing the default selector alone changes no row. - [ ] **Step 2: Run Vitest red** ```bash npx vitest run resources/js/composables/useActionPlanEditor.test.js ``` - [ ] **Step 3: Implement the pure editor contract** Keep client-only `id` out of payload. `updateStep` accepts an allow-listed patch: ```js const updateStep = (index, patch) => { const allowed = ['body', 'priority', 'action_type', 'assignee_uuid']; steps.value = steps.value.map((step, i) => i === index ? { ...step, ...Object.fromEntries(Object.entries(patch).filter(([key]) => allowed.includes(key))) } : step); }; ``` - [ ] **Step 4: Render row controls and AI-vs-human labels** Each row shows reason, suggested badge, confirmed priority, action type and assignee. Preserve approved plans as read-only. Approval disables until every non-empty row has valid selections. - [ ] **Step 5: Run JS tests/build, review and commit** ```bash npx vitest run resources/js/composables/useActionPlanEditor.test.js npm run build ``` Require Vue/accessibility review. Commit: ```bash git add resources/js/composables/useActionPlanEditor.js resources/js/composables/useActionPlanEditor.test.js resources/js/Components/Zoom/ActionPlanReviewModal.vue git commit -m "feat(zoom): add per-step action plan controls" ``` ### Task 6: Add proposal-only AI revisions **Files:** - Create: `resources/prompts/action_plan_revision.md` - Modify: `config/ai_prompts.php` - Modify: `src/Ai/AiRequest.php` - Create: `src/Zoom/Services/ActionPlanSuggestionContextBuilder.php` - Create: `app/Http/Requests/Manage/Zoom/ActionPlans/SuggestionRequest.php` - Create: `app/Http/Controllers/Manage/Zoom/ActionPlanSuggestionController.php` - Modify: `routes/web.php` - Create: `resources/js/Components/Zoom/ActionPlanRevisionPreview.vue` - Modify: `resources/js/Components/Zoom/ActionPlanReviewModal.vue` - Test: `tests/Feature/Zoom/ActionPlanSuggestionContextBuilderTest.php` - Test: `tests/Feature/Manage/Zoom/ActionPlanSuggestionTest.php` **Interfaces:** - POST body: `{instruction:string, items:[current HTTP item rows]}`. - JSON response: `{proposal:{items:[HTTP item rows]}, context:{transcript:{used,total,truncated}}}`. - No database write except normal `ai_requests` observability. - [ ] **Step 1: Write failing context, auth and no-write tests** Cover flag 404, non-Super Admin 403, bounded instruction/rows, transcript truncation honesty, prompt injection delimiter protection, invalid model JSON, unknown/duplicated keys, provider failure and database unchanged. - [ ] **Step 2: Verify failures** ```bash php artisan test tests/Feature/Zoom/ActionPlanSuggestionContextBuilderTest.php tests/Feature/Manage/Zoom/ActionPlanSuggestionTest.php ``` - [ ] **Step 3: Implement JSON-only proposal generation** Reuse `ZoomRecordingChatContextBuilder`; treat transcript, analysis, reviewer rows and instruction as quoted untrusted data. Prompt must return the complete proposed list and may only use submitted assignee UUIDs/enum codes. The server allow-lists known step keys and eligible assignees and rejects the entire malformed proposal with 422/503; it never saves. - [ ] **Step 4: Add explicit preview/apply UI** Show Current and Proposed. `Apply proposal` calls `editor.replaceWithProposal()` only. Closing/retrying leaves the editor unchanged. Save/Approve remain separate buttons. - [ ] **Step 5: Run tests/build, security review and commit** ```bash php artisan test tests/Feature/Zoom/ActionPlanSuggestionContextBuilderTest.php tests/Feature/Manage/Zoom/ActionPlanSuggestionTest.php npm run build ``` Commit: ```bash git add resources/prompts/action_plan_revision.md config/ai_prompts.php src/Ai/AiRequest.php src/Zoom/Services/ActionPlanSuggestionContextBuilder.php app/Http/Requests/Manage/Zoom/ActionPlans/SuggestionRequest.php app/Http/Controllers/Manage/Zoom/ActionPlanSuggestionController.php routes/web.php resources/js/Components/Zoom/ActionPlanRevisionPreview.vue resources/js/Components/Zoom/ActionPlanReviewModal.vue tests/Feature/Zoom/ActionPlanSuggestionContextBuilderTest.php tests/Feature/Manage/Zoom/ActionPlanSuggestionTest.php git commit -m "feat(zoom): propose reviewable action plan revisions" ``` ### Task 7: Preserve useful step-feedback evidence **Files:** - Create: `src/Conversation/Support/ConversationAnalysisHash.php` - Modify: `src/Lead/Repositories/LeadActionItemRepository.php` - Modify: `app/Http/Controllers/Manage/Leads/ActionItemsController.php` - Modify: `app/Http/Controllers/Manage/Zoom/ActionPlanFeedbackController.php` - Modify: `resources/js/Pages/Manage/Zoom/ActionPlans/Feedback.vue` - Test: `tests/Feature/Manage/Leads/ActionItemFeedbackTest.php` **Interfaces:** - Snapshot fields come from the server-loaded item/plan, never browser input. - `ConversationAnalysisHash::make(array $analysis): string` recursively key-sorts associative arrays, preserves list order, encodes with `JSON_UNESCAPED_UNICODE | JSON_UNESCAPED_SLASHES | JSON_THROW_ON_ERROR`, then returns SHA-256. - [ ] **Step 1: Write failing snapshot and authorization tests** Prove snapshot accuracy, re-rating refresh, source edit/delete survival, legacy null metadata, flag gate and assignee-only rating. - [ ] **Step 2: Verify failures** ```bash php artisan test tests/Feature/Manage/Leads/ActionItemFeedbackTest.php ``` - [ ] **Step 3: Implement server-authored snapshot upsert** Map fields under `lead_action_item_feedback` inside the controller/repository. Never accept snapshot fields from the Form Request. - [ ] **Step 4: Show snapshots and honest AI attribution** The admin listing shows rated text/priority/type even if the live item changed. If displaying an `ai_requests` row, label it “latest analysis request for this meeting”; do not call it an exact prompt version. - [ ] **Step 5: Test, review and commit** ```bash php artisan test tests/Feature/Manage/Leads/ActionItemFeedbackTest.php npm run build ``` ```bash git add src/Conversation/Support/ConversationAnalysisHash.php src/Lead/Repositories/LeadActionItemRepository.php app/Http/Controllers/Manage/Leads/ActionItemsController.php app/Http/Controllers/Manage/Zoom/ActionPlanFeedbackController.php resources/js/Pages/Manage/Zoom/ActionPlans/Feedback.vue tests/Feature/Manage/Leads/ActionItemFeedbackTest.php git commit -m "feat(leads): snapshot ai step feedback context" ``` ### Task 8: Turn the personal checklist into a role-aware workspace **Files:** - Create: `app/Http/Requests/Manage/Dashboard/TeamChecklistRequest.php` - Modify: `app/Http/Controllers/Manage/DashboardController.php` - Modify: `routes/web.php` - Modify: `resources/js/Pages/Manage/Dashboard.vue` - Modify: `resources/js/Components/TodaysChecklistModal.vue` - Modify: `resources/js/utils/actionPlanChecklist.js` - Modify: `resources/js/utils/actionPlanChecklist.test.js` - Test: `tests/Feature/Manage/DashboardChecklistTest.php` **Interfaces:** - Existing initial `actionPlanChecklist` remains the signed-in user's own open list. - New GET `/manage/dashboard/checklist/team` accepts `assignee`, `priority`, `cursor`; Super Admin only. - Team response: `{items:[{item,lead,plan,assignee,priority,action_type}],summary:{open,stranded},next_cursor}`. - [ ] **Step 1: Write failing own/team/auth/filter/order tests** Cover Sales isolation, Super Admin team access, non-Super 403, flag 404, Lead visibility, priority order, pagination, completed history and stranded rows that are counted without exposure to an ineligible old assignee. - [ ] **Step 2: Verify failures** ```bash php artisan test tests/Feature/Manage/DashboardChecklistTest.php ``` - [ ] **Step 3: Add priority payload and lazy Team query** Keep the landing query scoped to the current user. The team endpoint is separate, paginated and uses `LeadVisibility::apply()` plus Super Admin authorization. Order `priority ASC` (High=1) then plan approval/id descending. - [ ] **Step 4: Update copy and tabs** Visible title becomes `My Checklist`. Sales sees Current Tasks / Completed History. Super Admin additionally sees My Tasks / Team Tasks, with Team loaded only after selection. Show priority/type badges and assignee. - [ ] **Step 5: Run PHP/JS/build, review and commit** ```bash php artisan test tests/Feature/Manage/DashboardChecklistTest.php npx vitest run resources/js/utils/actionPlanChecklist.test.js npm run build ``` ```bash git add app/Http/Requests/Manage/Dashboard/TeamChecklistRequest.php app/Http/Controllers/Manage/DashboardController.php routes/web.php resources/js/Pages/Manage/Dashboard.vue resources/js/Components/TodaysChecklistModal.vue resources/js/utils/actionPlanChecklist.js resources/js/utils/actionPlanChecklist.test.js tests/Feature/Manage/DashboardChecklistTest.php git commit -m "feat(manage): add role-aware sales checklist workspace" ``` ### Task 9: Add safe contact actions for typed tasks **Files:** - Create: `resources/js/composables/useLeadContactActions.js` - Create: `resources/js/composables/useLeadContactActions.test.js` - Modify: `resources/js/Components/Sales/LeadCell.vue` - Modify: `resources/js/Components/TodaysChecklistModal.vue` - Modify: `app/Http/Controllers/Manage/DashboardController.php` - Test: `tests/Feature/Manage/DashboardChecklistTest.php` **Interfaces:** - `dial({leadUuid, phone, name})` logs with keepalive then sets `window.location.href = tel:`. - `whatsapp({phone})` opens normalized `wa.me` with `noopener`. - Checklist Lead payload includes `phone` only after Lead visibility has been applied. - [ ] **Step 1: Write failing JS and payload tests** Cover call log order, keepalive, cancel/no-call, WhatsApp digit normalization, invalid phone, action-type button visibility and no raw phone in unauthorized/team rows. - [ ] **Step 2: Run red tests** ```bash npx vitest run resources/js/composables/useLeadContactActions.test.js resources/js/Components/Sales/LeadCell.test.js php artisan test tests/Feature/Manage/DashboardChecklistTest.php ``` - [ ] **Step 3: Extract and reuse the contact handoff** Move the existing `LeadCell.vue` call-intent behavior without changing its observable contract. WhatsApp opens a new tab. Do not infer action type from task body. - [ ] **Step 4: Render task-specific controls and verify** Call button only for `TYPE_CALL`, WhatsApp only for `TYPE_WHATSAPP`, and neither when the phone is absent/invalid. - [ ] **Step 5: Review and commit** ```bash npx vitest run resources/js/composables/useLeadContactActions.test.js resources/js/Components/Sales/LeadCell.test.js php artisan test tests/Feature/Manage/DashboardChecklistTest.php npm run build ``` ```bash git add resources/js/composables/useLeadContactActions.js resources/js/composables/useLeadContactActions.test.js resources/js/Components/Sales/LeadCell.vue resources/js/Components/TodaysChecklistModal.vue app/Http/Controllers/Manage/DashboardController.php tests/Feature/Manage/DashboardChecklistTest.php git commit -m "feat(manage): add checklist contact actions" ``` ### Task 10: Complete workflow verification and local integration **Files:** - Modify only tests/docs needed to correct defects found during verification. - [ ] **Step 1: Run focused backend suite** ```bash php artisan test tests/Feature/Conversation/ConversationAnalyzerTest.php tests/Feature/Zoom/ZoomActionPlanFoundationTest.php tests/Feature/Zoom/ZoomActionPlanDraftTest.php tests/Feature/Zoom/ZoomActionPlanApprovalTest.php tests/Feature/Zoom/ActionPlanSuggestionContextBuilderTest.php tests/Feature/Manage/Zoom/ZoomActionPlansEndpointsTest.php tests/Feature/Manage/Zoom/ActionPlanSuggestionTest.php tests/Feature/Manage/Leads/ActionItemFeedbackTest.php tests/Feature/Manage/DashboardChecklistTest.php ``` Expected: all PASS. - [ ] **Step 2: Run frontend suite and builds** ```bash npm test npm run build ``` Expected: tests PASS and client + SSR builds succeed. Record unrelated baseline failures separately; do not label a new regression pre-existing without a clean-base reproduction. - [ ] **Step 3: Run formatting/static checks** Run Pint only on changed PHP files first, then the project's normal relevant static checks. Do not reformat unrelated files. - [ ] **Step 4: Rehearse the acceptance journey in `petav3.test`** Verify two different row assignees, priority sorting, revision preview/no-write, idempotent approval, Sales isolation, Team Tasks, call/WhatsApp handoff and feedback snapshots. Verify the flag-off state has no new UI/props and dedicated routes 404. - [ ] **Step 5: Final independent reviews** Run integration, security, database and Vue reviews over the complete diff. Fix every P0/P1 and justified P2 finding, rerun affected checks, and leave a concise evidence log. - [ ] **Step 6: Integrate locally** Ensure both worktrees are clean, create a recoverable backup ref for local `dev-chen`, then cherry-pick only this plan's task commits into the local `dev-chen` worktree. Do not push and do not open a PR.