ai-developer-skill-os 10.1.0 → 10.2.0

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 (98) hide show
  1. package/.agents/AGENTS.md +13 -0
  2. package/.agents/LICENSE +21 -21
  3. package/.agents/docs/ARCHITECTURE.md +56 -120
  4. package/.agents/docs/GOVERNANCE.md +3 -3
  5. package/.agents/docs/SPEC.md +137 -60
  6. package/.agents/docs/VERSIONING.md +25 -57
  7. package/.agents/docs/adr/0005-v10-platform-consolidation.md +50 -0
  8. package/.agents/docs/schemas/learning.schema.yml +22 -57
  9. package/.agents/docs/schemas/skill.schema.yml +116 -161
  10. package/.agents/docs/schemas/workflow.schema.yml +51 -26
  11. package/.agents/docs/skill-classification.md +1 -1
  12. package/.agents/registry/graph.json +171 -77
  13. package/.agents/registry/index.yaml +68 -35
  14. package/.agents/rules/coding.md +1 -1
  15. package/.agents/rules/command-safety.md +1 -1
  16. package/.agents/rules/global.md +1 -1
  17. package/.agents/rules/prompt-compiler.md +1 -1
  18. package/.agents/rules/safety.md +1 -1
  19. package/.agents/rules/security.md +1 -1
  20. package/.agents/rules/skill-quality.md +18 -3
  21. package/.agents/skills/_template/SKILL.md +238 -88
  22. package/.agents/skills/qk-api-data-discovery/SKILL.md +176 -190
  23. package/.agents/skills/qk-api-data-discovery/references/bronze-record-format.md +47 -0
  24. package/.agents/skills/qk-api-data-discovery/references/data-contract-yaml.md +72 -0
  25. package/.agents/skills/qk-api-data-discovery/references/discovery-report-template.md +110 -0
  26. package/.agents/skills/qk-backend-data/SKILL.md +150 -1
  27. package/.agents/skills/qk-bug-resolution/SKILL.md +153 -1
  28. package/.agents/skills/qk-bug-resolution/evals/scorecard.yaml +1 -1
  29. package/.agents/skills/qk-code-cleaner/SKILL.md +150 -1
  30. package/.agents/skills/qk-code-review/SKILL.md +195 -1
  31. package/.agents/skills/qk-code-review/evals/scorecard.yaml +1 -1
  32. package/.agents/skills/qk-code-review/references/ai/{v8-schema-validation.md → schema-validation.md} +2 -2
  33. package/.agents/skills/qk-code-review/references/cross-cutting/async-concurrency-patterns.md +515 -515
  34. package/.agents/skills/qk-code-review/references/cross-cutting/error-handling-principles.md +492 -492
  35. package/.agents/skills/qk-code-review/references/cross-cutting/n-plus-one-queries.md +309 -309
  36. package/.agents/skills/qk-code-review/references/cross-cutting/sql-injection-prevention.md +307 -307
  37. package/.agents/skills/qk-code-review/references/cross-cutting/xss-prevention.md +263 -263
  38. package/.agents/skills/qk-code-review/references/languages/angular.md +768 -768
  39. package/.agents/skills/qk-code-review/references/languages/c.md +890 -890
  40. package/.agents/skills/qk-code-review/references/languages/cpp.md +893 -893
  41. package/.agents/skills/qk-code-review/references/languages/css-less-sass.md +661 -661
  42. package/.agents/skills/qk-code-review/references/languages/django.md +985 -985
  43. package/.agents/skills/qk-code-review/references/languages/fastapi.md +580 -580
  44. package/.agents/skills/qk-code-review/references/languages/go.md +993 -993
  45. package/.agents/skills/qk-code-review/references/languages/java.md +409 -409
  46. package/.agents/skills/qk-code-review/references/languages/java8.md +586 -586
  47. package/.agents/skills/qk-code-review/references/languages/kotlin.md +1018 -1018
  48. package/.agents/skills/qk-code-review/references/languages/nestjs.md +593 -593
  49. package/.agents/skills/qk-code-review/references/languages/php.md +684 -684
  50. package/.agents/skills/qk-code-review/references/languages/python.md +1073 -1073
  51. package/.agents/skills/qk-code-review/references/languages/qt.md +757 -757
  52. package/.agents/skills/qk-code-review/references/languages/react.md +871 -871
  53. package/.agents/skills/qk-code-review/references/languages/ruby.md +964 -964
  54. package/.agents/skills/qk-code-review/references/languages/rust.md +846 -846
  55. package/.agents/skills/qk-code-review/references/languages/svelte.md +1064 -1064
  56. package/.agents/skills/qk-code-review/references/languages/swift.md +936 -936
  57. package/.agents/skills/qk-code-review/references/languages/typescript.md +1016 -1016
  58. package/.agents/skills/qk-code-review/references/languages/vue.md +924 -924
  59. package/.agents/skills/qk-code-review/references/languages/zig.md +440 -440
  60. package/.agents/skills/qk-devops-release/SKILL.md +150 -1
  61. package/.agents/skills/qk-feature-delivery/SKILL.md +153 -1
  62. package/.agents/skills/qk-feature-delivery/evals/scorecard.yaml +1 -1
  63. package/.agents/skills/qk-orchestrator/SKILL.md +187 -2
  64. package/.agents/skills/qk-orchestrator/evals/scorecard.yaml +1 -1
  65. package/.agents/skills/qk-orchestrator/references/routing-table.md +1 -1
  66. package/.agents/skills/qk-product-spec/SKILL.md +150 -1
  67. package/.agents/skills/qk-prompt-compiler/SKILL.md +188 -65
  68. package/.agents/skills/qk-ui-engineer/SKILL.md +150 -1
  69. package/.agents/workflows/_schema.yml +146 -146
  70. package/.agents/workflows/bug-resolution.yml +155 -121
  71. package/.agents/workflows/code-review.yml +127 -93
  72. package/.agents/workflows/context-discovery.yml +128 -94
  73. package/.agents/workflows/documentation.yml +124 -90
  74. package/.agents/workflows/feature-delivery.yml +158 -124
  75. package/.agents/workflows/production-release.yml +207 -173
  76. package/.agents/workflows/prompt-compilation.yml +126 -0
  77. package/.agents/workflows/refactor.yml +136 -102
  78. package/.agents/workflows/security-audit.yml +149 -115
  79. package/.agents/workflows/shared/quality-gate.yml +3 -1
  80. package/.agents/workflows/skin-governance.yml +149 -115
  81. package/.agents/workflows/spec-driven-development.yml +78 -71
  82. package/CHANGELOG.md +61 -0
  83. package/README.md +3 -3
  84. package/bin/install.js +38 -51
  85. package/package.json +68 -74
  86. package/tooling/build-registry.js +226 -208
  87. package/tooling/run-aar.js +55 -126
  88. package/tooling/sync-versions.js +2 -2
  89. package/tooling/validate-graph.js +100 -87
  90. package/tooling/validate-skills.js +32 -14
  91. package/.agents/docs/decisions/ADR-001-v8-migration.md +0 -58
  92. package/.agents/docs/decisions/ADR-002-workflow-separation.md +0 -50
  93. package/.agents/docs/decisions/ADR-003-registry-generated.md +0 -54
  94. package/.agents/docs/decisions/ADR-008-skill-boundary-review.md +0 -27
  95. package/.agents/registry/capability-graph.yml +0 -390
  96. package/.agents/registry/skills-index.yml +0 -305
  97. package/.agents/skills/_template/capability.yaml +0 -34
  98. package/.agents/skills/_template/evals/scorecard.yaml +0 -19
@@ -1,115 +1,149 @@
1
- name: skin-governance
2
- version: 9.2.0
3
- description: "Workflow tối ưu hóa, kiểm định và nâng cấp AI Skin — observe usage, measure success, improve, validate, promote."
4
-
5
- owner:
6
- type: engineering
7
- maintained_by: qk-team
8
-
9
- verification_profile: skill-evolution
10
-
11
- triggers:
12
- - "improve this skill"
13
- - "tối ưu skill"
14
- - "why does this skill keep failing"
15
- - "optimize this workflow"
16
- - "phân tích usage skill"
17
- - "nâng cấp skin"
18
- - "skin governance"
19
-
20
- ide_binding:
21
- slash_command: "/skin-tune"
22
- aliases: ["/evolve-skill", "/skin-governance"]
23
-
24
- steps:
25
- - id: observe_usage
26
- name: "Observe Usage"
27
- purpose: "Thống kê cách skill được gọi, token consumed, exit codes phân bổ"
28
- required: true
29
- inputs:
30
- - skill_logs
31
- - evals_traces
32
- outputs:
33
- - usage_metrics # trigger_rate, avg_tokens, exit_distribution
34
- actions:
35
- - "Đọc evals/traces/ của skill target"
36
- - "Thống kê trigger rate và token usage per exit code"
37
- - "Phát hiện pattern gọi sai skill (routing error)"
38
-
39
- - id: measure_success
40
- name: "Measure Success Rate"
41
- purpose: "Đánh giá chất lượng thực thi theo scorecard"
42
- required: true
43
- inputs:
44
- - usage_metrics
45
- - scorecard_yaml
46
- outputs:
47
- - success_rate # % SUCCESS / total runs
48
- - weak_areas # Steps hay fail nhất
49
- actions:
50
- - "Tính SUCCESS / FAILED / PARTIAL / BLOCKED ratio"
51
- - "Xác định bước nào trong workflow hay bị BLOCKED nhất"
52
- - "So sánh với threshold trong scorecard.yaml"
53
-
54
- - id: detect_failure
55
- name: "Detect Failure Patterns"
56
- purpose: "Tìm root cause của các lần fail — không đoán mò"
57
- required: true
58
- inputs:
59
- - execution_results
60
- - evals_traces
61
- outputs:
62
- - failure_patterns # Categorized: ambiguous_input | hallucination | budget_overrun | wrong_routing
63
- actions:
64
- - "Phân loại lỗi: ambiguous_input / hallucination / budget_overrun / wrong_routing"
65
- - "Tìm ví dụ cụ thể từ traces"
66
- - "Ghi root cause evidence cho từng category"
67
-
68
- - id: improve_skill
69
- name: "Propose Improvement"
70
- purpose: "Đề xuất thay đổi cụ thể — không thay đổi ngầm"
71
- required: true
72
- inputs:
73
- - failure_patterns
74
- - skill_body
75
- outputs:
76
- - improvement_proposal # Diff: what to add/remove/change + reason
77
- actions:
78
- - "Giảm noise trong instruction (xóa rule không được enforce)"
79
- - "Cập nhật examples với case thực tế từ traces"
80
- - "Tinh chỉnh token_budget nếu hay bị overrun"
81
- - "Cập nhật triggers nếu routing hay sai"
82
- - "KHÔNG tự sửa — trả proposal để human review"
83
- human_checkpoint:
84
- condition: always
85
- message: "Review improvement proposal trước khi apply vào SKILL.md"
86
-
87
- - id: validate
88
- name: "Validate Proposal"
89
- purpose: "Đảm bảo bản cải tiến không vi phạm governance rules"
90
- required: true
91
- inputs:
92
- - improvement_proposal
93
- outputs:
94
- - validation_result # PASS | FAIL với lý do
95
- actions:
96
- - "Kiểm tra schema compliance (SPEC.md frontmatter)"
97
- - "Đảm bảo không tạo scope overlap với skill khác"
98
- - "Verify triggers không trùng skill khác trong routing table"
99
-
100
- - id: promote
101
- name: "Promote to Stable"
102
- purpose: "Đưa bản cải tiến vào production sau khi đã validate"
103
- required: true
104
- inputs:
105
- - validation_result
106
- outputs:
107
- - promoted_skill # version bumped SKILL.md
108
- actions:
109
- - "Bump patch version (9.2.x → 9.2.x+1)"
110
- - "Cập nhật registry: node tooling/generate-registry.js"
111
- - "Ghi changelog entry"
112
-
113
- escalation:
114
- max_retries: 1
115
- on_failure: "Hủy bỏ bản cải tiến, giữ nguyên version hiện tại, ghi failure reason"
1
+ name: skin-governance
2
+ version: 10.2.0
3
+ platforms: [antigravity, claude, opencode]
4
+ runtime_version: 1
5
+ description: "Workflow tối ưu hóa, kiểm định và nâng cấp AI Skin — observe usage, measure success, improve, validate, promote."
6
+
7
+ owner:
8
+ type: engineering
9
+ maintained_by: qk-team
10
+
11
+ verification_profile: skill-evolution
12
+
13
+ triggers:
14
+ - "improve this skill"
15
+ - "tối ưu skill"
16
+ - "why does this skill keep failing"
17
+ - "optimize this workflow"
18
+ - "phân tích usage skill"
19
+ - "nâng cấp skin"
20
+ - "skin governance"
21
+
22
+ ide_binding:
23
+ slash_command: "/skin-tune"
24
+ aliases: ["/evolve-skill", "/skin-governance"]
25
+
26
+ steps:
27
+ - id: observe_usage
28
+ name: "Observe Usage"
29
+ purpose: "Thống kê cách skill được gọi, token consumed, exit codes phân bổ"
30
+ required: true
31
+ inputs:
32
+ - skill_logs
33
+ - evals_traces
34
+ outputs:
35
+ - usage_metrics # trigger_rate, avg_tokens, exit_distribution
36
+ actions:
37
+ - "Đọc evals/traces/ của skill target"
38
+ - "Thống kê trigger rate và token usage per exit code"
39
+ - "Phát hiện pattern gọi sai skill (routing error)"
40
+
41
+ - id: measure_success
42
+ name: "Measure Success Rate"
43
+ purpose: "Đánh giá chất lượng thực thi theo scorecard"
44
+ required: true
45
+ inputs:
46
+ - usage_metrics
47
+ - scorecard_yaml
48
+ outputs:
49
+ - success_rate # % SUCCESS / total runs
50
+ - weak_areas # Steps hay fail nhất
51
+ actions:
52
+ - "Tính SUCCESS / FAILED / PARTIAL / BLOCKED ratio"
53
+ - "Xác định bước nào trong workflow hay bị BLOCKED nhất"
54
+ - "So sánh với threshold trong scorecard.yaml"
55
+
56
+ - id: detect_failure
57
+ name: "Detect Failure Patterns"
58
+ purpose: "Tìm root cause của các lần fail — không đoán mò"
59
+ required: true
60
+ inputs:
61
+ - execution_results
62
+ - evals_traces
63
+ outputs:
64
+ - failure_patterns # Categorized: ambiguous_input | hallucination | budget_overrun | wrong_routing
65
+ actions:
66
+ - "Phân loại lỗi: ambiguous_input / hallucination / budget_overrun / wrong_routing"
67
+ - "Tìm ví dụ cụ thể từ traces"
68
+ - "Ghi root cause evidence cho từng category"
69
+
70
+ - id: improve_skill
71
+ name: "Propose Improvement"
72
+ purpose: "Đề xuất thay đổi cụ thể — không thay đổi ngầm"
73
+ required: true
74
+ inputs:
75
+ - failure_patterns
76
+ - skill_body
77
+ outputs:
78
+ - improvement_proposal # Diff: what to add/remove/change + reason
79
+ actions:
80
+ - "Giảm noise trong instruction (xóa rule không được enforce)"
81
+ - "Cập nhật examples với case thực tế từ traces"
82
+ - "Tinh chỉnh token_budget nếu hay bị overrun"
83
+ - "Cập nhật triggers nếu routing hay sai"
84
+ - "KHÔNG tự sửa — trả proposal để human review"
85
+ human_checkpoint:
86
+ condition: always
87
+ message: "Review improvement proposal trước khi apply vào SKILL.md"
88
+
89
+ - id: validate
90
+ name: "Validate Proposal"
91
+ purpose: "Đảm bảo bản cải tiến không vi phạm governance rules"
92
+ required: true
93
+ inputs:
94
+ - improvement_proposal
95
+ outputs:
96
+ - validation_result # PASS | FAIL với lý do
97
+ actions:
98
+ - "Kiểm tra schema compliance (SPEC.md frontmatter)"
99
+ - "Đảm bảo không tạo scope overlap với skill khác"
100
+ - "Verify triggers không trùng skill khác trong routing table"
101
+
102
+ - id: promote
103
+ name: "Promote to Stable"
104
+ purpose: "Đưa bản cải tiến vào production sau khi đã validate"
105
+ required: true
106
+ inputs:
107
+ - validation_result
108
+ outputs:
109
+ - promoted_skill # version bumped SKILL.md
110
+ actions:
111
+ - "Bump patch version (10.2.x → 10.2.x+1)"
112
+ - "Cập nhật registry: node tooling/build-registry.js"
113
+ - "Ghi changelog entry"
114
+
115
+ escalation:
116
+ max_retries: 1
117
+ on_failure: "Hủy bỏ bản cải tiến, giữ nguyên version hiện tại, ghi failure reason"
118
+
119
+ # ── V10.2 Additions ──────────────────────────────
120
+ # Confidence Model
121
+ # | Level | Condition | Action |
122
+ # |--------|------------------------|--------------|
123
+ # | HIGH | Step inputs provided | Proceed |
124
+ # | MEDIUM | Some assumptions needed | Note assumptions |
125
+ # | LOW | Insufficient evidence | EXIT: BLOCKED|
126
+ #
127
+ # Exit Codes
128
+ # | Code | Meaning | When |
129
+ # |---------|-------------------------------|-------------------------|
130
+ # | SUCCESS | Step completed and verified | All criteria met |
131
+ # | PARTIAL | Step done with minor gaps | Some checks skipped |
132
+ # | BLOCKED | Missing precondition | Ask user |
133
+ # | FAILED | Step failed after max retries | Report error |
134
+ #
135
+ # Evidence Format:
136
+ # [SEVERITY] step_id: [step-name]
137
+ # Reason: [why this matters]
138
+ # Confidence: [HIGH|MEDIUM|LOW]
139
+ # Fix: [suggestion]
140
+ #
141
+ # Compliance:
142
+ # Runtime Standard: 11/11
143
+ # Frontmatter Complete: yes
144
+ # References Valid: yes
145
+ # Decision Trees: PASS
146
+ # Thresholds Defined: PASS
147
+ # schema_version: 10.2.0
148
+ # runtime_version: 1
149
+ # platforms: [antigravity, claude, opencode]
@@ -1,6 +1,8 @@
1
1
  name: spec-driven-development
2
- version: 9.3.0
3
- description: "Workflow phát triển sản phẩm dựa trên đặc tả kỹ thuật (Specification-Driven Development): Idea → Requirements → Acceptance Criteria (BDD) → Technical Spec → Plan → Verify."
2
+ version: 10.2.0
3
+ platforms: [antigravity, claude, opencode]
4
+ runtime_version: 1
5
+ description: "Pipeline phát triển dựa trên đặc tả: Requirement → Acceptance Criteria → Technical Spec → Implementation."
4
6
 
5
7
  owner:
6
8
  type: engineering
@@ -9,101 +11,106 @@ owner:
9
11
  verification_profile: spec-driven
10
12
 
11
13
  triggers:
12
- - "write a spec"
13
- - "viết đặc tả"
14
- - "spec this out"
14
+ - "viết spec"
15
+ - "viết prd"
15
16
  - "acceptance criteria"
16
- - "prd"
17
- - "technical spec"
18
- - "user story"
19
- - "design before implementing"
17
+ - "bdd"
20
18
 
21
19
  ide_binding:
22
20
  slash_command: "/spec"
23
- aliases: ["/grill-me", "/prd", "/spec-driven"]
21
+ aliases: ["/prd", "/specify"]
24
22
 
25
23
  steps:
26
- - id: clarify_requirements
27
- name: "Làm rõ yêu cầu & Bóc tách ý tưởng"
28
- purpose: "Biến ý tưởng thô sơ thành mục tiêu sản phẩm rõ ràng, không mơ hồ"
24
+ - id: triage
25
+ name: "Triage & Clarify"
26
+ purpose: "Xác định quy mô task và kích hoạt /grill-me nếu thiếu thông tin"
29
27
  required: true
30
- inputs:
31
- - user_idea
32
- - dev_profile
33
28
  outputs:
34
- - problem_statement # Vấn đề cần giải quyết
35
- - target_actors # Ai sử dụng tính năng này
36
- - scope_boundary # In-scope vs Out-of-scope rõ ràng
29
+ - task_scale
30
+ - clarity_level
37
31
  actions:
38
- - "Gợi ý người dùng chạy lệnh slash /grill-me để phỏng vấn sâu nếu yêu cầu còn nhiều giả định chưa rõ"
39
- - "Xác định rõ vấn đề cốt lõi (Core Problem) và giá trị kinh doanh (Value)"
40
- - "Vạch rõ ranh giới: Làm gì (In-scope) và KHÔNG làm gì (Out-of-scope)"
41
- exit_on_fail: BLOCKED
32
+ - "Xác định Nhỏ/Vừa/Lớn"
33
+ - "Kích hoạt /grill-me nếu thiếu thông tin"
42
34
 
43
- - id: acceptance_criteria
44
- name: "Xây dựng Acceptance Criteria (BDD)"
45
- purpose: "Định nghĩa định lượng tiêu chí hoàn thành theo chuẩn Given / When / Then"
35
+ - id: gap_scan
36
+ name: "Codebase Gap Scan"
37
+ purpose: "Đối chiếu models & APIs hiện có để đánh giá khả thi"
46
38
  required: true
47
- inputs:
48
- - problem_statement
49
- - target_actors
50
- - scope_boundary
51
39
  outputs:
52
- - criteria_matrix # Bảng Given / When / Then cho từng use-case
53
- - edge_cases # Kịch bản biên, lỗi, rớt mạng, mất quyền
40
+ - reusable_components
41
+ - new_requirements
54
42
  actions:
55
- - "Viết kịch bản theo chuẩn BDD (Given / When / Then) cho từng User Story"
56
- - "Liệt kê ít nhất 3 edge cases (dữ liệu rỗng, timeout, quyền truy cập không hợp lệ)"
57
- - "Xác định rõ HTTP status code mong đợi cho từng kịch bản lỗi"
43
+ - "Quét codebase cho models, APIs hiện có"
44
+ - "Xác định Gap Analysis"
58
45
 
59
- - id: technical_specification
60
- name: "Thiết kế Đặc tả Kỹ thuật (Technical Spec)"
61
- purpose: "Xác định kiến trúc hệ thống, API contracts, và Data schema trước khi code"
46
+ - id: bdd_write
47
+ name: "BDD Spec Writing"
48
+ purpose: "Viết Acceptance Criteria chuẩn Gherkin"
62
49
  required: true
63
- inputs:
64
- - criteria_matrix
65
- - edge_cases
66
50
  outputs:
67
- - data_models # Entities, Interfaces, DB Schema migrations
68
- - api_contracts # Endpoint, Method, Request/Response JSON schemas
69
- - architecture_diagram # Sơ đồ luồng ASCII Unicode Box Drawing
51
+ - bdd_scenarios
52
+ - acceptance_criteria
70
53
  actions:
71
- - "Vẽ sơ đồ luồng dữ liệu (Data Flow) bằng Unicode/ASCII Box Drawing (chuẩn GEMINI.md)"
72
- - "Thiết kế Type contracts (TypeScript interfaces / Pydantic models / Protobuf)"
73
- - "Xác định rõ các điểm chốt chặn bảo mật (Boundary Validation R-SEC-04)"
74
- - "Đảm bảo contract giữa Frontend và Backend đồng bộ 100%"
54
+ - "Viết Given/When/Then cho Happy Path + Edge Cases"
55
+ human_checkpoint:
56
+ condition: "Task scale = Lớn"
57
+ message: "Feature lớn. Xác nhận BDD spec trước khi tiếp tục."
75
58
 
76
- - id: create_plan
77
- name: "Lập Kế hoạch Triển khai (Implementation Plan)"
78
- purpose: "Phân rã spec thành các task cụ thể và kích hoạt phê duyệt tương tác"
59
+ - id: plan
60
+ name: "Implementation Plan"
61
+ purpose: "Tạo implementation_plan.md và kích hoạt modal"
79
62
  required: true
80
- inputs:
81
- - technical_specification
82
63
  outputs:
83
- - implementation_plan # File implementation_plan.md
64
+ - implementation_plan_path
84
65
  actions:
85
- - "Phân chia task theo thứ tự an toàn: Types/Schema → Business Logic/Service → UI/Route → Tests"
86
- - "Ước lượng rủi ro và blast radius cho từng file"
87
- - "Trên Antigravity IDE: Tạo implementation_plan.md kèm RequestFeedback: true để kích hoạt Modal phê duyệt"
66
+ - "Tạo implementation_plan.md với RequestFeedback: true"
88
67
  human_checkpoint:
89
- condition: "Task có độ phức tạp medium/high (≥ 2 files hoặc sửa schema DB)"
90
- message: "Bản đặc tả và kế hoạch kỹ thuật đã sẵn sàng. Xác nhận để chuyển sang bước thực thi."
68
+ condition: "Always"
69
+ message: "Review implementation_plan trước khi code."
91
70
 
92
- - id: verify_compliance
93
- name: "Kiểm định Tuân thủ Đặc tả"
94
- purpose: "Đối chiếu mã nguồn sau khi code với 100% Acceptance Criteria ban đầu"
71
+ - id: verify
72
+ name: "Verify Against Spec"
73
+ purpose: "Xác nhận implementation match spec"
95
74
  required: true
96
- inputs:
97
- - source_code
98
- - criteria_matrix
99
75
  outputs:
100
- - compliance_report # Pass rate, tiêu chí chưa đạt (nếu có)
76
+ - spec_compliance
101
77
  actions:
102
- - "Kiểm tra từng tiêu chí Given/When/Then — đánh dấu ✅ hoặc ❌"
103
- - "Chạy automated tests nếu có để chứng minh tiêu chí đã đạt"
104
- - "Không cho phép đóng task nếu còn bất kỳ acceptance criterion nào chưa thỏa mãn"
78
+ - "Đối chiếu từng acceptance criteria"
79
+ - "Verify edge cases covered"
105
80
  exit_on_fail: PARTIAL
106
81
 
107
82
  escalation:
108
- max_retries: 2
109
- on_failure: "Báo cáo độ lệch so với spec, chỉ ra blocker và yêu cầu chỉ thị từ Tech Lead / User"
83
+ max_retries: 1
84
+ on_failure: "Report spec compliance gaps and ask user to update spec"
85
+
86
+ # ── V10.2 Additions ──────────────────────────────
87
+ # Confidence Model
88
+ # | Level | Condition | Action |
89
+ # |--------|------------------------|--------------|
90
+ # | HIGH | Step inputs provided | Proceed |
91
+ # | MEDIUM | Some assumptions needed | Note assumptions |
92
+ # | LOW | Insufficient evidence | EXIT: BLOCKED|
93
+ #
94
+ # Exit Codes
95
+ # | Code | Meaning | When |
96
+ # |---------|-------------------------------|-------------------------|
97
+ # | SUCCESS | Step completed and verified | All criteria met |
98
+ # | PARTIAL | Step done with minor gaps | Some checks skipped |
99
+ # | BLOCKED | Missing precondition | Ask user |
100
+ # | FAILED | Step failed after max retries | Report error |
101
+ #
102
+ # Evidence Format:
103
+ # [SEVERITY] step_id: [step-name]
104
+ # Reason: [why this matters]
105
+ # Confidence: [HIGH|MEDIUM|LOW]
106
+ # Fix: [suggestion]
107
+ #
108
+ # Compliance:
109
+ # Runtime Standard: 11/11
110
+ # Frontmatter Complete: yes
111
+ # References Valid: yes
112
+ # Decision Trees: PASS
113
+ # Thresholds Defined: PASS
114
+ # schema_version: 10.2.0
115
+ # runtime_version: 1
116
+ # platforms: [antigravity, claude, opencode]
package/CHANGELOG.md CHANGED
@@ -5,6 +5,26 @@ Format: [Keep a Changelog](https://keepachangelog.com/en/1.0.0/)
5
5
 
6
6
  ---
7
7
 
8
+ ## [10.2.0] — 2026-09-24 — AI Developer Skill OS v10.2: 3 IDEs (Antigravity, Claude, OpenCode), Platform-Specific Sections, graph.json edges
9
+
10
+ ### Added
11
+ - 3 IDE support: Antigravity, Claude, OpenCode
12
+ - Platform-Specific Instructions section in every SKILL.md
13
+ - `platforms` and `runtime_version` fields in all frontmatters
14
+ - `.claude/CLAUDE.md`, `.opencode/config.yaml` created
15
+ - CI workflow updated to ubuntu-latest
16
+
17
+ ### Changed
18
+ - Removed Cursor, Windsurf, Kilo, Codex support
19
+ - Removed old `/skills/` directory and `capability-graph.yml`
20
+ - Removed stale worktree `.kilo/worktrees/stealth-octagon`
21
+
22
+ ### Fixed
23
+ - `copyOpenCodeConfig` order fix in install.js
24
+ - `validate-skills.js` V10 status default
25
+ - `build-registry.js` dependency mapping with crossEdges
26
+ - `sync-versions.js` V9→V10 fix
27
+
8
28
  ## [10.1.0] — 2026-09-15 — AI Developer Skill OS v10.1: 11 Super-Skills, API-to-Data Discovery & Prompt Compiler v11.1
9
29
 
10
30
  ### Added
@@ -228,3 +248,44 @@ Format: [Keep a Changelog](https://keepachangelog.com/en/1.0.0/)
228
248
  ## [7.0.0] — 2026
229
249
 
230
250
  - V7 release: 22 skills, YAML frontmatter, token budget system
251
+
252
+ ---
253
+
254
+ ## [10.2.0] — 2026-09-24 — AI Developer Skill OS v10.2: Clean Architecture & Full Validation
255
+
256
+ ### Upgraded
257
+ - All 11 SKILL.md files version bumped to 10.2.0
258
+ - All 9 rule files version bumped to 10.2.0
259
+ - All 12 workflow YAML files version bumped to 10.2.0
260
+ - _template/SKILL.md upgraded from V9.2 to V10.2
261
+
262
+ ### Added — Every Skill Now Has:
263
+ - **Confidence Model** — HIGH/MEDIUM/LOW routing confidence
264
+ - **Exit Codes** — SUCCESS/PARTIAL/BLOCKED/FAILED
265
+ - **Evidence Format** — Structured evidence with severity + confidence
266
+ - **Compliance** — schema_version 10.2.0, runtime_version 1
267
+
268
+ ### Enhanced
269
+ - **qk-orchestrator**: Added Role Adaptation, Failure Path, Retry Policy (108→187 lines)
270
+ - **qk-code-review**: Added Role Adaptation, Failure Path, Evidence Format (132→220 lines)
271
+ - **qk-api-data-discovery**: Reduced 470→330 lines, split report templates to references/
272
+ - **qk-prompt-compiler**: Added Core Principles, Role Adaptation, Evidence Gate (321→408 lines)
273
+ - **7 other skills**: Added missing Exit Codes + Confidence Model + Evidence Format
274
+
275
+ ### Removed
276
+ - Deleted old `/skills/` directory (22-file legacy system)
277
+ - Deleted `capability-graph.yml` and `skills-index.yml` (old registry)
278
+ - Deleted `.kilo/worktrees/stealth-octagon` (stale git worktree)
279
+ - Deleted `_template/capability.yaml` (old system manifest)
280
+ - Deleted `_template/evals/scorecard.yaml` (old V8 format)
281
+
282
+ ### Fixed Tooling
283
+ - `validate-skills.js`: Updated to check V10 sections instead of V8 fields
284
+ - `build-registry.js`: Updated from V8.2 to V10.2, added edges array to graph.json
285
+ - `validate-graph.js`: Updated to read graph.json, fixed orphan detection
286
+ - `sync-versions.js`: Updated targetVersion from 9.1.0 to 10.2.0
287
+
288
+ ### Validation
289
+ - `npm run test:registry`: 10/10 tests passed
290
+ - `node tooling/validate-skills.js`: 11/11 valid
291
+ - `node tooling/validate-graph.js`: **Score 100/100** (0 orphans, 0 cycles, 11 edges)
package/README.md CHANGED
@@ -1,8 +1,8 @@
1
- # AI Developer Skill OS v10.1.0 — Lean Core Edition
1
+ # AI Developer Skill OS v10.2.0 — Lean Core Edition
2
2
 
3
- > **v10.1.0 "Lean Core Revolution & API-to-Data Discovery Integration"**
3
+ > **v10.2.0 "Lean Core Revolution & API-to-Data Discovery Integration"**
4
4
 
5
- Phiên bản **v10.1.0** giải quyết triệt để vấn đề phân mảnh quá mức (extreme fragmentation) từ 37 thư mục vi kỹ năng (micro-skills) xuống còn **11 Core Super-Skills** mạnh mẽ, không làm mất bất kỳ tính năng chuyên sâu nào, đồng thời tích hợp toàn diện cỗ máy biên dịch Prompt Compiler v11.1 và Discovery Engine cho Antigravity IDE.
5
+ Phiên bản **v10.2.0** giải quyết triệt để vấn đề phân mảnh quá mức (extreme fragmentation) từ 37 thư mục vi kỹ năng (micro-skills) xuống còn **11 Core Super-Skills** mạnh mẽ, không làm mất bất kỳ tính năng chuyên sâu nào, đồng thời tích hợp toàn diện cỗ máy biên dịch Prompt Compiler v10.2 và Discovery Engine cho Antigravity IDE.
6
6
 
7
7
  ---
8
8