@dzhechkov/keysarium 1.7.15 → 1.7.19

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 (83) hide show
  1. package/.dz-manifest.json +214 -54
  2. package/README.md +29 -6
  3. package/README.ru.md +17 -0
  4. package/package.json +3 -3
  5. package/sbom.json +455 -55
  6. package/src/cli.js +32 -2
  7. package/src/commands/doctor.js +25 -1
  8. package/src/commands/init.js +56 -12
  9. package/src/commands/update.js +30 -3
  10. package/src/utils.js +7 -0
  11. package/templates/.agents/skills/casarium/SKILL.md +29 -0
  12. package/templates/.claude/agents/code-goal-planner.md +447 -0
  13. package/templates/.claude/commands/casarium.md +3 -1
  14. package/templates/.claude/commands/cjm-prototype.md +2 -2
  15. package/templates/.claude/commands/research.md +4 -1
  16. package/templates/.claude/rules/checkpoint-protocol.md +2 -1
  17. package/templates/.claude/rules/research-quality.md +11 -6
  18. package/templates/.claude/shards/phase-2-research.shard.md +12 -8
  19. package/templates/.claude/shards/phase-25-cjm.shard.md +5 -3
  20. package/templates/.claude/skills/feature-adr/SKILL.md +17 -15
  21. package/templates/.claude/skills/feature-adr/modules/03-adr.md +1 -1
  22. package/templates/.claude/skills/feature-adr/modules/03.5-ideation-swarm.md +1 -1
  23. package/templates/.claude/skills/feature-adr/modules/06-implementation-plan.md +2 -2
  24. package/templates/.claude/skills/feature-adr/modules/08-qe.md +1 -1
  25. package/templates/.claude/skills/feature-adr/modules/09-fleet-qe.md +12 -12
  26. package/templates/.claude/skills/feature-adr/references/agentic-qe/LICENSE.agentic-qe +21 -0
  27. package/templates/.claude/skills/feature-adr/references/agentic-qe/README.md +20 -16
  28. package/templates/.claude/skills/feature-adr/references/agentic-qe/UPSTREAM-MANIFEST.json +246 -0
  29. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-brutal-honesty-review/README.md +218 -0
  30. package/templates/.claude/skills/feature-adr/references/agentic-qe/{brutal-honesty-review.md → qe-ref-brutal-honesty-review/REFERENCE.md} +14 -1
  31. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-brutal-honesty-review/resources/assessment-rubrics.md +295 -0
  32. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-brutal-honesty-review/resources/review-template.md +102 -0
  33. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-brutal-honesty-review/schemas/output.json +291 -0
  34. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-brutal-honesty-review/scripts/assess-code.sh +179 -0
  35. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-brutal-honesty-review/scripts/assess-tests.sh +223 -0
  36. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-brutal-honesty-review/scripts/validate-config.json +34 -0
  37. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-enterprise-integration-testing/REFERENCE.md +758 -0
  38. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-enterprise-integration-testing/evals/enterprise-integration-testing.yaml +158 -0
  39. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-enterprise-integration-testing/schemas/output.json +74 -0
  40. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-enterprise-integration-testing/scripts/validate-config.json +25 -0
  41. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-qcsd-ideation-swarm/REFERENCE.md +226 -0
  42. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-qcsd-ideation-swarm/evals/qcsd-ideation-swarm.yaml +138 -0
  43. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-qcsd-ideation-swarm/schemas/output.json +568 -0
  44. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-qcsd-ideation-swarm/scripts/validate-config.json +25 -0
  45. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-qcsd-ideation-swarm/steps/01-flag-detection.md +57 -0
  46. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-qcsd-ideation-swarm/steps/02-core-agents.md +29 -0
  47. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-qcsd-ideation-swarm/steps/03-batch1-results.md +15 -0
  48. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-qcsd-ideation-swarm/steps/04-conditional-agents.md +23 -0
  49. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-qcsd-ideation-swarm/steps/05-decision-synthesis.md +29 -0
  50. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-qcsd-ideation-swarm/steps/06-report-generation.md +18 -0
  51. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-qcsd-ideation-swarm/steps/07-learning-persistence.md +29 -0
  52. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-qcsd-ideation-swarm/steps/08-final-output.md +18 -0
  53. package/templates/.claude/skills/feature-adr/references/agentic-qe/{qe-coverage-analysis.md → qe-ref-qe-coverage-analysis/REFERENCE.md} +42 -3
  54. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-qe-coverage-analysis/evals/qe-coverage-analysis.yaml +494 -0
  55. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-qe-coverage-analysis/schemas/output.json +286 -0
  56. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-qe-coverage-analysis/scripts/validate-config.json +42 -0
  57. package/templates/.claude/skills/feature-adr/references/agentic-qe/{qe-requirements-validation.md → qe-ref-qe-requirements-validation/REFERENCE.md} +13 -3
  58. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-qe-requirements-validation/evals/qe-requirements-validation.yaml +598 -0
  59. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-qe-requirements-validation/schemas/output.json +587 -0
  60. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-qe-requirements-validation/scripts/validate-config.json +37 -0
  61. package/templates/.claude/skills/feature-adr/references/agentic-qe/{regression-testing.md → qe-ref-regression-testing/REFERENCE.md} +30 -49
  62. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-regression-testing/evals/regression-testing.yaml +504 -0
  63. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-regression-testing/schemas/output.json +311 -0
  64. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-regression-testing/scripts/validate-config.json +50 -0
  65. package/templates/.claude/skills/feature-adr/references/agentic-qe/{risk-based-testing.md → qe-ref-risk-based-testing/REFERENCE.md} +23 -52
  66. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-risk-based-testing/evals/risk-based-testing.yaml +141 -0
  67. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-risk-based-testing/schemas/output.json +480 -0
  68. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-risk-based-testing/scripts/validate-config.json +25 -0
  69. package/templates/.claude/skills/feature-adr/references/agentic-qe/{shift-left-testing.md → qe-ref-shift-left-testing/REFERENCE.md} +33 -88
  70. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-shift-left-testing/evals/shift-left-testing.yaml +145 -0
  71. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-shift-left-testing/schemas/output.json +459 -0
  72. package/templates/.claude/skills/feature-adr/references/agentic-qe/qe-ref-shift-left-testing/scripts/validate-config.json +25 -0
  73. package/templates/CLAUDE.md +3 -1
  74. package/templates/docs/feature-adr/04-admin-guide.md +1 -1
  75. package/templates/docs/improvement-directions.md +3 -3
  76. package/templates/docs/keysarium/02-admin-guide.md +2 -2
  77. package/templates/docs/keysarium/03-user-guide.md +3 -3
  78. package/templates/docs/keysarium/05-architecture.md +9 -8
  79. package/templates/docs/keysarium/06-user-flows.md +4 -4
  80. package/templates/lib/platform-templates/copilot.md +1 -1
  81. package/templates/lib/platform-templates/opencode.md +1 -1
  82. package/templates/.claude/skills/feature-adr/references/agentic-qe/enterprise-integration-testing.md +0 -111
  83. package/templates/.claude/skills/feature-adr/references/agentic-qe/qcsd-ideation-swarm.md +0 -126
@@ -401,7 +401,7 @@ CHECKPOINT X: [Phase Name] Complete
401
401
  | `domain-specific.md` | Доменные правила (банк, ритейл, enterprise, healthcare) |
402
402
  | `file-conventions.md` | Конвенции именования файлов и директорий |
403
403
  | `modular-reuse.md` | Правила модульного переиспользования скиллов и команд |
404
- | `research-quality.md` | Требования к качеству research (PARANOID mode) |
404
+ | `research-quality.md` | Выбор режима GOAP и требования к источникам |
405
405
  | `witness-chain.md` | Witness chain integration: SHA-256 хеширование артефактов |
406
406
  | `reward-learning.md` | Reward-calibrated learning: memory_query/memory_store протокол |
407
407
  | `background-workers.md` | Background workers: isolation, model routing, concurrency limits |
@@ -700,7 +700,7 @@ Markdown-документ с чёткой структурой:
700
700
 
701
701
  ## Instructions (Variant D — Trend Research)
702
702
  1. Загрузить goap-research-ed25519 SKILL.md
703
- 2. Исследовать 5 категорий трендов (PARANOID mode)
703
+ 2. Исследовать 5 категорий трендов в режиме, разрешённом возможностями источников (`moderate` по умолчанию)
704
704
  3. Синтезировать Future-Ready CJM
705
705
  4. Записать в промежуточный файл: _variant_d_draft.md
706
706
 
@@ -628,7 +628,7 @@ CHECKPOINT 1: Explore Complete
628
628
  Даже в Quick Mode:
629
629
  - Phase 2.5 (CJM Prototype) **не пропускается** (только упрощается)
630
630
  - `08_executive_summary.md` **обязателен**
631
- - Research в PARANOID mode -- **все утверждения с источниками**
631
+ - Research с указанием источников; обычные источники работают в GOAP `moderate`
632
632
  - Checkpoint-ы **сохраняются** (но можно отвечать `ок` быстрее)
633
633
 
634
634
  ### 5.4. Распределение времени в Quick Mode (1.5 часа)
@@ -946,9 +946,9 @@ A: Домен определяется автоматически по ключ
946
946
 
947
947
  ---
948
948
 
949
- **Q: Что такое PARANOID mode в research?**
949
+ **Q: Когда доступен `paranoid` в research?**
950
950
 
951
- A: Это режим повышенной верификации данных. Каждое фактическое утверждение должно иметь проверяемый источник. Утверждения без источников помечаются как `[UNVERIFIED]`. Это защищает от галлюцинаций AI и обеспечивает доверие жюри.
951
+ A: Только когда для выбранного класса источников есть закреплённые ключи издателей и подписанная выдача. Обычные публикации и сайты используют `moderate`: факты снабжаются ссылками, пробелы помечаются `[UNVERIFIED]`. Ссылка сама по себе не доказывает подпись издателя и не гарантирует истинность утверждения.
952
952
 
953
953
  ---
954
954
 
@@ -212,7 +212,7 @@ Read(".claude/skills/<skill-name>/SKILL.md")
212
212
  | `domain-specific.md` | Доменные правила (банки, ритейл, enterprise, healthcare) |
213
213
  | `file-conventions.md` | Соглашения об именовании файлов и структуре директорий |
214
214
  | `modular-reuse.md` | Правила модульного переиспользования компонентов |
215
- | `research-quality.md` | PARANOID mode: требования к верификации источников |
215
+ | `research-quality.md` | Выбор режима GOAP и требования к источникам |
216
216
 
217
217
  ---
218
218
 
@@ -313,15 +313,16 @@ researches/<case-slug>/
313
313
  - **Оптимизация моделей:** haiku для простых операций, sonnet для анализа, opus для креативной работы
314
314
  - **Синтез:** после каждого параллельного этапа оркестратор объединяет результаты
315
315
 
316
- ### 4. Верифицируемость: PARANOID mode для research
316
+ ### 4. Верифицируемость: режим по возможностям источников
317
317
 
318
- Все исследовательские данные проходят строгую верификацию:
318
+ Для обычных источников доступен GOAP `moderate`. `strict` и `paranoid` требуют
319
+ реальных закреплённых ключей издателей и подписанной выдачи выбранного класса
320
+ источников. Во всех режимах действуют требования к качеству:
319
321
 
320
- - Каждое утверждение должно иметь проверяемый источник
322
+ - Существенные фактические утверждения снабжаются источниками
321
323
  - Нулевая толерантность к галлюцинированным цитатам
322
- - Порог уверенности: минимум 0.99
323
324
  - Непроверенные данные маркируются как `[UNVERIFIED]`
324
- - Перед завершением фазы -- чеклист верификации
325
+ - Перед завершением фазы -- чеклист доказательств; `GOAL_UNREACHABLE` блокирует checkpoint
325
326
 
326
327
  ### 5. Расширяемость: легко добавить новый домен или фазу
327
328
 
@@ -410,7 +411,7 @@ sequenceDiagram
410
411
  O->>P2: Загрузить goap-research-ed25519 SKILL.md
411
412
  Note over P2: 3 агента параллельно
412
413
  P2->>P2: Аналоги || Технологии || Регуляторика
413
- Note over P2: PARANOID mode: верификация каждого утверждения
414
+ Note over P2: GOAP moderate для обычных источников; ссылки и пробелы явны
414
415
  P2-->>O: 02_research_findings.md
415
416
  O->>U: Checkpoint 2
416
417
  U->>O: "ок"
@@ -499,7 +500,7 @@ graph TB
499
500
  subgraph "Phase 2.5: CJM Prototype (3 агента)"
500
501
  P25_O["Orchestrator Phase 2.5"] --> P25_A1["Agent 1:<br/>CJM Variant A"]
501
502
  P25_O --> P25_A2["Agent 2:<br/>CJM Variant B + C"]
502
- P25_O --> P25_A3["Agent 3:<br/>Trend Research (Variant D)<br/>GOAP PARANOID"]
503
+ P25_O --> P25_A3["Agent 3:<br/>Trend Research (Variant D)<br/>GOAP mode by capabilities"]
503
504
  P25_A1 --> P25_S["Синтез → 4 варианта<br/>prototype/cjm-prototype.jsx"]
504
505
  P25_A2 --> P25_S
505
506
  P25_A3 --> P25_S
@@ -66,7 +66,7 @@ sequenceDiagram
66
66
 
67
67
  Note over C: ═══ Phase 2: Research ═══
68
68
  C->>FS: Read .claude/skills/goap-research-ed25519/SKILL.md
69
- C->>A: Spawn 3 агента (PARANOID mode)
69
+ C->>A: Spawn 3 агента (режим по доступным проверкам источников)
70
70
  A->>A: Agent 1: Аналоги + метрики
71
71
  A->>A: Agent 2: Технологии + Anti-patterns
72
72
  A->>A: Agent 3: Регуляторика + market data
@@ -154,7 +154,7 @@ User: /casarium Кейс: Автоматизация контакт-центра
154
154
  - Checkpoint 1 -> пользователь подтверждает "ок"
155
155
 
156
156
  **Шаг 5. Phase 2 -- Research**
157
- - Загружается скилл `goap-research-ed25519` в PARANOID mode
157
+ - Загружается скилл `goap-research-ed25519`; обычные источники исследуются в `moderate`
158
158
  - 3 агента параллельно: аналоги, технологии, регуляторика
159
159
  - Синтез с верификацией каждого утверждения
160
160
  - Результат: `02_research_findings.md`
@@ -264,7 +264,7 @@ sequenceDiagram
264
264
 
265
265
  - Phase 2.5 (CJM) все равно **не пропускается** -- упрощается до 2 вариантов
266
266
  - `08_executive_summary.md` все равно **обязателен**
267
- - PARANOID mode для research **сохраняется** (лучше меньше утверждений, но верифицированных)
267
+ - Требование указывать источники и пробелы **сохраняется**; режим GOAP выбирается по возможностям источников
268
268
  - Checkpoints **сохраняются** (но можно пройти быстрее)
269
269
 
270
270
  ---
@@ -865,7 +865,7 @@ flowchart TD
865
865
 
866
866
  **Принципы обработки:**
867
867
  - Research **никогда не пропускается** -- даже при отсутствии данных фиксируется сам факт отсутствия
868
- - PARANOID mode: лучше `[NO_DATA]`, чем непроверенное утверждение
868
+ - Лучше `[NO_DATA]`, чем неподтверждённое утверждение, представленное как факт
869
869
  - Пробелы маркируются явно -- это ценная информация для последующих фаз
870
870
  - Пользователь всегда информируется о состоянии данных на checkpoint
871
871
 
@@ -34,7 +34,7 @@ Create `.github/copilot-instructions.md` with the following structure:
34
34
  {Summarize the rule to 5-10 key bullet points.
35
35
  Focus on actionable instructions.
36
36
  Omit implementation details specific to Claude Code.
37
- Preserve critical constraints (PARANOID mode, mandatory phases, etc.)
37
+ Preserve critical constraints (research mode capability gate, mandatory phases, etc.)
38
38
  }
39
39
 
40
40
  ---
@@ -24,7 +24,7 @@ project:
24
24
  instructions: |
25
25
  {Extract the critical rules from CLAUDE.md:
26
26
  - File conventions
27
- - Research quality requirements (PARANOID mode)
27
+ - Research quality requirements and GOAP mode capability gate
28
28
  - Checkpoint protocol summary
29
29
  - Anti-patterns to avoid
30
30
  Keep to 30-50 lines.}
@@ -1,111 +0,0 @@
1
- # Enterprise Integration Testing (Condensed for Feature ADR)
2
-
3
- > Source: agentic-qe `enterprise-integration-testing` skill (Tier 3)
4
- > Condensed from 735 lines — SAP-specific sections removed, universal patterns retained.
5
-
6
- ## Purpose
7
-
8
- Verify that independently correct modules/systems work correctly together.
9
- The biggest risks are in the seams between components, not within them.
10
-
11
- ## Integration Testing Protocol
12
-
13
- ### 1. Map E2E Flow
14
-
15
- Before writing tests, map the complete flow:
16
- ```
17
- User Action → API → Service A → Service B → Database → Response
18
- ```
19
-
20
- Identify every integration boundary (API calls, DB queries, message queues, file I/O).
21
-
22
- ### 2. Test Each Boundary
23
-
24
- For each integration point:
25
-
26
- | Boundary Type | What to Test |
27
- |---|---|
28
- | API (REST/GraphQL) | Contract compliance, schema validation, error codes |
29
- | Database | Query correctness, transaction isolation, migration compatibility |
30
- | Message Queue | Message format, ordering, DLQ handling, retry logic |
31
- | File I/O | Format validation, encoding, path handling |
32
- | External Service | Mock fidelity, timeout handling, circuit breaker |
33
-
34
- ### 3. Cross-Module Data Consistency
35
-
36
- After each integration event, verify:
37
- - Data written by module A is correctly read by module B
38
- - No data loss or transformation errors at boundaries
39
- - Eventual consistency windows are respected (async flows)
40
-
41
- ### 4. Contract Testing
42
-
43
- ```
44
- Consumer defines expected contract
45
- → Provider validates against contract
46
- → Break = blocked merge
47
- ```
48
-
49
- Key checks:
50
- - Request/response schemas match between caller and callee
51
- - Error response formats are consistent
52
- - API versioning is respected
53
- - No breaking changes in shared interfaces
54
-
55
- ### 5. Error Recovery
56
-
57
- Test failure scenarios at every boundary:
58
- - What happens when Service B is down?
59
- - Are partial operations compensated/rolled back?
60
- - Do correlation IDs propagate through the chain?
61
- - Are DLQ messages reprocessable?
62
-
63
- ## Integration Quality Gates
64
-
65
- | Gate | Threshold | Action |
66
- |---|---|---|
67
- | Cross-module data consistency | 100% for critical paths | BLOCK on failure |
68
- | API contract compliance | 100% | BLOCK on failure |
69
- | Error recovery | All compensation paths tested | WARN if partial |
70
- | Performance | E2E flow < defined SLA | WARN if exceeded |
71
-
72
- ## Hierarchical Testing Strategy
73
-
74
- Execute integration tests in phases:
75
- 1. **Contract validation** — verify interfaces match
76
- 2. **Integration testing** — verify boundary behavior
77
- 3. **E2E validation** — verify complete flows
78
- 4. **Authorization check** — verify access control across modules
79
-
80
- ## QCSD Flags for Integration
81
-
82
- ```
83
- HAS_CROSS_MODULE_DATA: true → Data replicated/shared across modules
84
- HAS_ASYNC_PROCESSING: true → Message queues, event-driven flows
85
- HAS_EXTERNAL_SERVICES: true → Third-party API integrations
86
- HAS_DATABASE_SHARING: true → Multiple modules access same DB
87
- ```
88
-
89
- ## Best Practices
90
-
91
- ### Do
92
- - Map the full E2E flow before writing any tests
93
- - Test each integration boundary separately AND end-to-end
94
- - Validate data consistency across all modules after each event
95
- - Use service virtualization for hard-to-provision dependencies
96
- - Include correlation ID tracing in test assertions
97
-
98
- ### Avoid
99
- - Testing only through the UI without API-level tests
100
- - Ignoring async processing errors (DLQ, failed events)
101
- - Assuming cross-module data is immediately consistent
102
- - Testing integrations only in isolation (mock everything)
103
- - Deploying without verifying no breaking API changes
104
-
105
- ## Agent Coordination
106
-
107
- When used as Fleet QE Agent 3 (Integration Integrity Checker):
108
- - Focus scope: cross-module interactions within the feature
109
- - Input: `{CODE_CHANGES}` file list, `{ARCHITECTURE}` component diagram
110
- - Output: integration issues found, contract violations, data consistency risks
111
- - Verdict per issue: BLOCKER / WARNING / INFO
@@ -1,126 +0,0 @@
1
- ---
2
- name: qcsd-ideation-swarm
3
- description: "QCSD Ideation phase swarm for Quality Criteria sessions using HTSM v6.3, Risk Storming, and Testability analysis before development begins."
4
- source: https://github.com/proffesor-for-testing/agentic-qe/tree/main/assets/skills/qcsd-ideation-swarm
5
- version: 7.5.1
6
- trust_tier: 3
7
- agents:
8
- core: [qe-quality-criteria-recommender, qe-risk-assessor, qe-requirements-validator]
9
- conditional: [qe-accessibility-auditor, qe-security-auditor, qe-qx-partner]
10
- ---
11
-
12
- # QCSD Ideation Swarm (Adapted for Feature-ADR)
13
-
14
- Shift-left quality engineering swarm applied to feature ADR and requirements analysis.
15
-
16
- ---
17
-
18
- ## Phase 1: Flag Detection
19
-
20
- Scan feature requirements/ADR and SET these flags:
21
-
22
- ```
23
- □ HAS_UI = FALSE
24
- Set TRUE if feature contains: UI, frontend, visual, component, screen, page, form, button,
25
- modal, dialog, dashboard, widget, interface, display, view, layout, CSS, styling
26
-
27
- □ HAS_SECURITY = FALSE
28
- Set TRUE if feature contains: auth, security, credential, token, encrypt, PII, compliance,
29
- password, login, session, OAuth, JWT, permission, role, access control, RBAC, sensitive
30
-
31
- □ HAS_UX = FALSE
32
- Set TRUE if feature contains: user experience, UX, journey, usability, user flow, persona,
33
- friction, delight, onboarding, retention, engagement
34
- ```
35
-
36
- ## Phase 2: Spawn Core Agents (3 parallel)
37
-
38
- ALL THREE core agents MUST be spawned in a single message:
39
-
40
- ### Agent 1: Quality Criteria Recommender (HTSM v6.3)
41
- Analyze ALL 10 HTSM categories with weight and testability score:
42
- 1. Capability — does it do what's required?
43
- 2. Reliability — under stress, load, edge cases
44
- 3. Usability — learnability, accessibility, efficiency
45
- 4. Charisma — visual design, responsiveness
46
- 5. Security — authentication, authorization, data protection
47
- 6. Scalability — growth handling, resource limits
48
- 7. Compatibility — browsers, devices, APIs, integrations
49
- 8. Performance — speed, throughput, resource usage
50
- 9. Installability — deployment, configuration, migration
51
- 10. Development — testability, maintainability, portability
52
-
53
- ### Agent 2: Risk Assessor (SFDIPOT)
54
- Apply SFDIPOT framework:
55
- - **S**tructure: Code architecture, component dependencies
56
- - **F**unction: Business logic, calculations, workflows
57
- - **D**ata: Data integrity, migrations, edge values
58
- - **I**nterfaces: API contracts, integration points, UI interactions
59
- - **P**latform: OS, browser, infrastructure dependencies
60
- - **O**perations: Deployment, monitoring, maintenance
61
- - **T**ime: Scheduling, concurrency, timeouts, caching
62
-
63
- Identify minimum 10 risks with probability (1-5), impact (1-5), and risk score.
64
-
65
- ### Agent 3: Requirements Validator (Testability)
66
- Apply 10 Principles of Testability, score each 0-100:
67
- 1. Observability — can we see what the system does?
68
- 2. Controllability — can we put it in the state we need?
69
- 3. Decomposability — can we test parts independently?
70
- 4. Simplicity — is it only as complex as needed?
71
- 5. Stability — does it change predictably?
72
- 6. Information — do we have what we need to test?
73
- 7. Communication — do we know what to test?
74
- 8. Relation to prior work — can we leverage past tests?
75
- 9. Time — do we have enough time to test?
76
- 10. Accountability — who owns quality?
77
-
78
- ## Phase 3: Conditional Agents (based on flags)
79
-
80
- | Flag | Agent | Analysis |
81
- |------|-------|----------|
82
- | HAS_UI | qe-accessibility-auditor | WCAG 2.2 AA compliance assessment |
83
- | HAS_SECURITY | qe-security-auditor | STRIDE threat modeling |
84
- | HAS_UX | qe-qx-partner | User journey and experience quality analysis |
85
-
86
- ## Phase 4: GO/NO-GO Decision
87
-
88
- | Metric | NO-GO | CONDITIONAL | GO |
89
- |--------|-------|-------------|-----|
90
- | Testability Score | < 40 | 40-79 | ≥ 80 |
91
- | HTSM Coverage | < 6/10 | 6-7/10 | ≥ 8/10 |
92
- | AC Completeness | < 50% | 50-89% | ≥ 90% |
93
- | Critical Risks | > 2 | 1-2 | 0 |
94
-
95
- **NO-GO**: Stop. Requirements need rework before development.
96
- **CONDITIONAL**: Proceed with advisory warnings. Include findings in judge context.
97
- **GO**: Full confidence. Proceed to implementation planning.
98
-
99
- ## Phase 5: Synthesis Report
100
-
101
- Generate consolidated report with:
102
- 1. Executive summary (GO/CONDITIONAL/NO-GO verdict)
103
- 2. Quality criteria scores (10 HTSM categories)
104
- 3. Risk matrix (top 10 risks by score)
105
- 4. Testability assessment (10 principles)
106
- 5. Test ideas (categorized by type: unit/integration/E2E)
107
- 6. Recommendations (prioritized by risk × impact)
108
-
109
- ## Integration with Feature-ADR
110
-
111
- This swarm runs as Step 3.5 in the feature-adr pipeline:
112
- - **Input**: Refined ADR from Step 3 + Requirements from Step 1
113
- - **Output**: Quality assessment report in `features/<slug>/03.5_ideation_report.md`
114
- - **Gate**: NO-GO blocks progression to Step 6 (Implementation Plan)
115
- - **CONDITIONAL**: Warnings are passed to Steps 6-8 as context
116
-
117
- ## Enforcement Rules
118
-
119
- | Rule | Requirement |
120
- |------|------------|
121
- | E1 | ALL THREE core agents must be spawned. No exceptions. |
122
- | E2 | All parallel agents in a SINGLE message. |
123
- | E3 | STOP and WAIT after each batch. No proceeding early. |
124
- | E4 | Conditional agents MUST be spawned if flags are TRUE. |
125
- | E5 | GO/CONDITIONAL/NO-GO logic applied exactly as specified. |
126
- | E6 | Full report structure generated. No abbreviated versions. |