adi_dev_workflow 1.0.0 → 1.1.1

package/frameworks/skills/sdd-task-plan-expert-workspace/iteration-2/eval-1-happy-path/without_skill/outputs/transcript.md

@@ -0,0 +1,45 @@
+# Transcript — Tool Calls
+
+## 1. Read
+- **File:** `docs/feature-user/v1/spec_tech.md` (lines 1-200)
+- **Purpose:** Read the approved SPEC_TECH document (first section)
+
+## 2. Read (overflow)
+- **File:** Persisted output from step 1 (full file was too large, 49.8KB)
+- **Purpose:** Attempt to read full output — still too large (56.1KB)
+
+## 3. Read
+- **File:** `docs/feature-user/v1/spec_tech.md` (lines 200-600)
+- **Purpose:** Read sections 6-14 of the SPEC_TECH (business rules, flows, API contracts, persistence, security, error handling, config, logging, test strategy)
+
+## 4. Read
+- **File:** `docs/feature-user/v1/spec_tech.md` (lines 600-919)
+- **Purpose:** Read remaining sections of the SPEC_TECH (test scenarios, traceability, files involved, final checklist)
+
+## 5. Bash
+- **Command:** `ls internal/` — List internal directory structure
+- **Purpose:** Understand existing codebase layout
+
+## 6. Bash
+- **Command:** `ls internal/db/migrations/` — List existing migrations
+- **Purpose:** Check if migration 002 already exists (it does)
+
+## 7. Bash
+- **Command:** `ls -la .claude/skills/sdd-task-plan-expert-workspace/iteration-2/eval-1-happy-path/without_skill/outputs/`
+- **Purpose:** Check if output directory exists
+
+## 8. Bash
+- **Command:** `mkdir -p` for output directory
+- **Purpose:** Ensure output directory exists
+
+## 9. Read
+- **File:** Existing `response.md` in output directory
+- **Purpose:** Required before overwriting with Write tool
+
+## 10. Write
+- **File:** `.claude/skills/sdd-task-plan-expert-workspace/iteration-2/eval-1-happy-path/without_skill/outputs/response.md`
+- **Purpose:** Save the complete TASK PLAN response
+
+## 11. Write
+- **File:** `.claude/skills/sdd-task-plan-expert-workspace/iteration-2/eval-1-happy-path/without_skill/outputs/transcript.md`
+- **Purpose:** Save this transcript

package/frameworks/skills/sdd-task-plan-expert-workspace/iteration-2/eval-1-happy-path/without_skill/timing.json

@@ -0,0 +1 @@
+{"total_tokens": 69814, "duration_ms": 347289, "total_duration_seconds": 347.3}

package/frameworks/skills/sdd-task-plan-expert-workspace/iteration-2/eval-2-spec-simples/eval_metadata.json

@@ -0,0 +1,12 @@
+{
+  "eval_id": 2,
+  "eval_name": "spec-tech-simples-alteracao-pontual",
+  "prompt": "Preciso de um TASK PLAN para esta feature simples. O SPEC_TECH aprovado esta abaixo:\n\n---\n# SPEC_TECH — Adicionar Telefone ao Usuario v1\n\n## 1. Identificacao\n| Campo | Valor |\n|-------|-------|\n| **Feature/Projeto** | Adicionar Telefone ao Modulo de Usuario |\n| **Versao** | v1 |\n| **Data** | 2026-03-07 |\n| **PRD** | docs/feature-telefone-usuario/v1/prd.md |\n\n## 2. Resumo Tecnico\nAdicionar campo `telefone` (opcional) a tabela `usuarios` e expor via gRPC nos endpoints CreateUser e UpdateUser.\n\n## 3. Definicoes Tecnicas\n\n### 3.1 Banco de Dados\n- Nova migracao: `ALTER TABLE usuarios ADD COLUMN telefone TEXT;`\n\n### 3.2 SQLC\n- Atualizar queries CreateUser e UpdateUser para incluir coluna `telefone`\n\n### 3.3 Repository\n- Adicionar campo `Phone` ao modelo User\n- Mapear `telefone` (banco) para `Phone` (Go)\n\n### 3.4 Service\n- Aceitar campo opcional phone em CreateUser e UpdateUser\n- Validacao: se informado, deve ter entre 10 e 15 caracteres\n\n### 3.5 Handler gRPC\n- Adicionar campo `phone` nas mensagens proto CreateUserRequest, UpdateUserRequest e User\n\n### 3.6 Proto\n- Atualizar user.proto com campo `string phone = X;`\n\n## 4. User Stories Relacionadas\n- US-10: Como usuario, quero informar meu telefone para receber notificacoes por SMS\n---",
+  "assertions": [
+    {"id": "b1", "text": "Gera um numero proporcional de tasks (entre 3 e 6) — nao over-engineers com tasks demais para uma alteracao simples", "type": "llm_judge"},
+    {"id": "b2", "text": "Respeita a ordem de dependencias das camadas (migracao > SQLC > repository > service > handler > proto)", "type": "llm_judge"},
+    {"id": "b3", "text": "Mapeia a US-10 na tabela de rastreabilidade do task_plan", "type": "llm_judge"},
+    {"id": "b4", "text": "Cada task segue o template completo (secoes 1-8) ou marca secoes N/A com justificativa", "type": "llm_judge"},
+    {"id": "b5", "text": "Salva task_plan.md como documento de REFERENCIA (sem corpo detalhado de tasks dentro dele)", "type": "llm_judge"}
+  ]
+}

package/frameworks/skills/sdd-task-plan-expert-workspace/iteration-2/eval-2-spec-simples/with_skill/grading.json

@@ -0,0 +1,32 @@
+{
+  "eval_id": 2,
+  "eval_name": "spec-tech-simples-alteracao-pontual",
+  "configuration": "with_skill",
+  "expectations": [
+    {
+      "text": "Gera um numero proporcional de tasks (entre 3 e 6) — nao over-engineers com tasks demais para uma alteracao simples",
+      "passed": true,
+      "evidence": "Gerou 6 tasks (T1 migracao, T2 SQLC, T3 repository, T4 service, T5 proto+handler, T6 validacao). Proporcional a complexidade."
+    },
+    {
+      "text": "Respeita a ordem de dependencias das camadas (migracao > SQLC > repository > service > handler > proto)",
+      "passed": true,
+      "evidence": "Ordem: T1(migracao) -> T2(SQLC) -> T3(repository) -> T4(service) -> T5(proto+handler) -> T6(validacao). Dependencias respeitadas."
+    },
+    {
+      "text": "Mapeia a US-10 na tabela de rastreabilidade do task_plan",
+      "passed": true,
+      "evidence": "Secao 5 do task_plan.md: 'US-10: Informar telefone para receber SMS' mapeada para T1-T6. Cada task individual tambem referencia 'User Stories Relacionadas: US-10'."
+    },
+    {
+      "text": "Cada task segue o template completo (secoes 1-8) ou marca secoes N/A com justificativa",
+      "passed": true,
+      "evidence": "Todas as 6 tasks possuem secoes: 1.Identificacao, 2.Objetivo, 3.Descricao Detalhada, 4.Aceite Tecnico, 5.Arquivos Impactados, 6.Testes, 7.Notas, 8.Checklist Final."
+    },
+    {
+      "text": "Salva task_plan.md como documento de REFERENCIA (sem corpo detalhado de tasks dentro dele)",
+      "passed": true,
+      "evidence": "task_plan.md aparece como documento separado ao final com secoes: Identificacao, Objetivo, Macro-Fases, Lista de Tasks (links para TN.md), Rastreabilidade, Dependencias, Criterios, Riscos, Checklist."
+    }
+  ]
+}