@su-record/vibe 2.8.0 → 2.8.1
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.
- package/README.md +275 -333
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -6,9 +6,9 @@
|
|
|
6
6
|
[](https://www.typescriptlang.org/)
|
|
7
7
|
[](https://opensource.org/licenses/MIT)
|
|
8
8
|
|
|
9
|
-
|
|
9
|
+
**One install adds 56 agents, 36 skills, multi-LLM orchestration, and automated quality gates to your AI coding workflow.**
|
|
10
10
|
|
|
11
|
-
Claude Code, Codex, Cursor, Gemini CLI
|
|
11
|
+
Works with Claude Code, Codex, Cursor, and Gemini CLI.
|
|
12
12
|
|
|
13
13
|
```bash
|
|
14
14
|
npm install -g @su-record/vibe
|
|
@@ -17,408 +17,378 @@ vibe init
|
|
|
17
17
|
|
|
18
18
|
---
|
|
19
19
|
|
|
20
|
-
##
|
|
20
|
+
## Why Vibe
|
|
21
21
|
|
|
22
|
-
AI
|
|
23
|
-
Vibe
|
|
22
|
+
AI generates working code, but quality is left to chance.
|
|
23
|
+
Vibe solves this structurally.
|
|
24
24
|
|
|
25
|
-
|
|
|
26
|
-
|
|
27
|
-
| AI
|
|
28
|
-
|
|
|
29
|
-
|
|
|
30
|
-
| AI
|
|
31
|
-
|
|
|
32
|
-
|
|
|
25
|
+
| Problem | Solution |
|
|
26
|
+
|---------|----------|
|
|
27
|
+
| AI scatters `any` types | Quality Gate blocks `any` / `@ts-ignore` |
|
|
28
|
+
| Expecting one-shot perfection | SPEC → Implement → Verify staged workflow |
|
|
29
|
+
| Merging without review | 12 agents run parallel code review |
|
|
30
|
+
| Accepting AI output blindly | GPT + Gemini cross-validation |
|
|
31
|
+
| Losing context between sessions | Session RAG auto-saves and restores |
|
|
32
|
+
| Getting lost on complex tasks | SwarmOrchestrator auto-decomposes + parallelizes |
|
|
33
33
|
|
|
34
|
-
###
|
|
34
|
+
### Design Philosophy
|
|
35
35
|
|
|
36
|
-
|
|
|
37
|
-
|
|
38
|
-
| **Easy Vibe Coding** |
|
|
39
|
-
| **Minimum Quality Guaranteed** |
|
|
40
|
-
| **Iterative Reasoning** |
|
|
36
|
+
| Principle | Description |
|
|
37
|
+
|-----------|-------------|
|
|
38
|
+
| **Easy Vibe Coding** | Fast flow — think collaboratively with AI |
|
|
39
|
+
| **Minimum Quality Guaranteed** | Type safety, code quality, security — automatic baseline |
|
|
40
|
+
| **Iterative Reasoning** | Break down problems, ask questions, reason together |
|
|
41
41
|
|
|
42
42
|
---
|
|
43
43
|
|
|
44
|
-
##
|
|
44
|
+
## Workflow
|
|
45
45
|
|
|
46
|
-
|
|
47
|
-
|
|
48
|
-
|
|
49
|
-
|
|
50
|
-
|
|
51
|
-
|
|
52
|
-
|
|
53
|
-
### Codex 플러그인 통합
|
|
54
|
-
|
|
55
|
-
Codex Claude Code 플러그인(`codex-plugin-cc`) 설치 시 워크플로우 전 단계에서 자동 활용:
|
|
56
|
-
|
|
57
|
-
| 워크플로우 | Codex 활용 | 명령 |
|
|
58
|
-
|-----------|-----------|------|
|
|
59
|
-
| **spec review** | SPEC 설계 도전 리뷰 | `/codex:adversarial-review` |
|
|
60
|
-
| **run** | 병렬 구현 위임 | `/codex:rescue --background` |
|
|
61
|
-
| **run / review** | 3중 교차 코드 리뷰 (GPT+Gemini+Codex) | `/codex:review` |
|
|
62
|
-
| **run / review** | auto-fix 실패 시 폴백 | `/codex:rescue` |
|
|
63
|
-
| **verify** | 최종 리뷰 게이트 | `/codex:review` |
|
|
64
|
-
| **Stop 훅** | 코드 변경 시 자동 리뷰 | `codex-review-gate.js` |
|
|
46
|
+
```mermaid
|
|
47
|
+
flowchart LR
|
|
48
|
+
A["/vibe.spec"] --> B["/vibe.spec.review"]
|
|
49
|
+
B --> C["/vibe.run"]
|
|
50
|
+
C --> D["Auto Review"]
|
|
51
|
+
D --> E["Done"]
|
|
52
|
+
```
|
|
65
53
|
|
|
66
|
-
|
|
54
|
+
1. **`/vibe.spec`** — Define requirements as a SPEC document (GPT + Gemini parallel research)
|
|
55
|
+
2. **`/vibe.spec.review`** — SPEC quality review + Codex adversarial review (triple cross-validation)
|
|
56
|
+
3. **`/vibe.run`** — Implement from SPEC (Codex rescue parallel delegation) + triple code review
|
|
57
|
+
4. **Auto Review** — 12 specialized agents review in parallel + boundary verification, P1/P2 auto-fix
|
|
67
58
|
|
|
68
|
-
|
|
59
|
+
Add `ultrawork` to automate the entire pipeline:
|
|
69
60
|
|
|
70
|
-
```
|
|
71
|
-
|
|
72
|
-
├── .codex-plugin/plugin.json # 플러그인 매니페스트
|
|
73
|
-
├── agents/ # 56개 에이전트 (순수 마크다운)
|
|
74
|
-
├── skills/ # 전역 스킬
|
|
75
|
-
└── AGENTS.md # VIBE 지시사항
|
|
61
|
+
```bash
|
|
62
|
+
/vibe.run "feature" ultrawork
|
|
76
63
|
```
|
|
77
64
|
|
|
78
65
|
---
|
|
79
66
|
|
|
80
|
-
##
|
|
67
|
+
## Multi-CLI Support
|
|
81
68
|
|
|
82
|
-
|
|
83
|
-
|
|
84
|
-
|
|
85
|
-
|
|
86
|
-
|
|
87
|
-
|
|
88
|
-
```
|
|
69
|
+
| CLI | Install Location | Agents | Skills | Instructions |
|
|
70
|
+
|-----|-----------------|--------|--------|-------------|
|
|
71
|
+
| **Claude Code** | `~/.claude/agents/` | 56 | `~/.claude/skills/` | `CLAUDE.md` |
|
|
72
|
+
| **Codex** | `~/.codex/plugins/vibe/` | 56 | Plugin built-in | `AGENTS.md` |
|
|
73
|
+
| **Cursor** | `~/.cursor/agents/` | 56 | `~/.cursor/skills/` | `.cursorrules` |
|
|
74
|
+
| **Gemini CLI** | `~/.gemini/agents/` | 56 | `~/.gemini/skills/` | `GEMINI.md` |
|
|
89
75
|
|
|
90
|
-
|
|
91
|
-
2. **`/vibe.spec.review`** — SPEC 품질 리뷰 + Codex adversarial review (3중 교차 검증)
|
|
92
|
-
3. **`/vibe.run`** — SPEC 기반 구현 (Codex rescue 병렬 위임) + 3중 코드 리뷰
|
|
93
|
-
4. **자동 리뷰** — 12개 전문 에이전트 병렬 검토 + Codex/경계면 검증, P1/P2 자동 수정
|
|
76
|
+
### Codex Plugin Integration
|
|
94
77
|
|
|
95
|
-
`
|
|
78
|
+
When the Codex Claude Code plugin (`codex-plugin-cc`) is installed, Vibe automatically integrates it across every workflow stage:
|
|
96
79
|
|
|
97
|
-
|
|
98
|
-
|
|
99
|
-
|
|
80
|
+
| Workflow | Codex Usage | Command |
|
|
81
|
+
|----------|------------|---------|
|
|
82
|
+
| **spec review** | Adversarial SPEC challenge | `/codex:adversarial-review` |
|
|
83
|
+
| **run** | Parallel implementation delegation | `/codex:rescue --background` |
|
|
84
|
+
| **run / review** | Triple code review (GPT + Gemini + Codex) | `/codex:review` |
|
|
85
|
+
| **run / review** | Fallback on auto-fix failure | `/codex:rescue` |
|
|
86
|
+
| **verify** | Final review gate | `/codex:review` |
|
|
87
|
+
| **Stop hook** | Auto-review on code changes | `codex-review-gate.js` |
|
|
88
|
+
|
|
89
|
+
Auto-skips when Codex is not installed — existing workflow continues as-is.
|
|
100
90
|
|
|
101
91
|
---
|
|
102
92
|
|
|
103
|
-
##
|
|
93
|
+
## Agents (56)
|
|
104
94
|
|
|
105
|
-
###
|
|
95
|
+
### Core Agents (19)
|
|
106
96
|
|
|
107
|
-
|
|
|
108
|
-
|
|
109
|
-
|
|
|
110
|
-
|
|
|
111
|
-
|
|
|
112
|
-
|
|
|
97
|
+
| Category | Agents |
|
|
98
|
+
|----------|--------|
|
|
99
|
+
| **Exploration** | Explorer (High / Medium / Low) |
|
|
100
|
+
| **Implementation** | Implementer (High / Medium / Low) |
|
|
101
|
+
| **Architecture** | Architect (High / Medium / Low) |
|
|
102
|
+
| **Utility** | Searcher, Tester, Simplifier, Refactor Cleaner, Build Error Resolver, Compounder, Diagrammer, E2E Tester, UI Previewer, Junior Mentor |
|
|
113
103
|
|
|
114
|
-
###
|
|
104
|
+
### Review Agents (12)
|
|
115
105
|
|
|
116
106
|
Security, Performance, Architecture, Complexity, Simplicity, Data Integrity, Test Coverage, Git History, TypeScript, Python, Rails, React
|
|
117
107
|
|
|
118
|
-
### UI/UX
|
|
108
|
+
### UI/UX Agents (8)
|
|
119
109
|
|
|
120
|
-
|
|
110
|
+
Design intelligence backed by 48 CSV datasets. Industry analysis → Design system generation → Implementation guide → Accessibility audit.
|
|
121
111
|
|
|
122
|
-
|
|
|
123
|
-
|
|
112
|
+
| Phase | Agents |
|
|
113
|
+
|-------|--------|
|
|
124
114
|
| SPEC | ui-industry-analyzer, ui-design-system-gen, ui-layout-architect |
|
|
125
115
|
| RUN | ui-stack-implementer, ui-dataviz-advisor |
|
|
126
116
|
| REVIEW | ux-compliance-reviewer, ui-a11y-auditor, ui-antipattern-detector |
|
|
127
117
|
|
|
128
|
-
### QA &
|
|
118
|
+
### QA & Research (11)
|
|
129
119
|
|
|
130
|
-
|
|
|
131
|
-
|
|
120
|
+
| Category | Agents |
|
|
121
|
+
|----------|--------|
|
|
132
122
|
| **QA** | QA Coordinator, Edge Case Finder, Acceptance Tester |
|
|
133
|
-
|
|
|
134
|
-
|
|
|
123
|
+
| **Research** | Best Practices, Framework Docs, Codebase Patterns, Security Advisory |
|
|
124
|
+
| **Analysis** | Requirements Analyst, UX Advisor, API Documenter, Changelog Writer |
|
|
135
125
|
|
|
136
|
-
QA Coordinator
|
|
126
|
+
QA Coordinator analyzes changed code and dispatches appropriate QA agents in parallel, then produces a unified QA report.
|
|
137
127
|
|
|
138
|
-
###
|
|
128
|
+
### Event Agents (6)
|
|
139
129
|
|
|
140
130
|
Event Content, Event Image, Event Speaker, Event Ops, Event Comms, Event Scheduler
|
|
141
131
|
|
|
142
132
|
---
|
|
143
133
|
|
|
144
|
-
##
|
|
134
|
+
## Skills (36)
|
|
145
135
|
|
|
146
|
-
|
|
147
|
-
|-----------|------|------|------|
|
|
148
|
-
| **Claude** | Opus / Sonnet / Haiku | SPEC 작성, 코드 리뷰, 오케스트레이션 | 내장 (Claude Code) |
|
|
149
|
-
| **GPT** | gpt-5.4 | 추론, 아키텍처, 엣지케이스 분석 | Codex CLI / API Key |
|
|
150
|
-
| **Codex** | gpt-5.3-codex | 구현, 멀티파일 코딩 | Codex CLI / Plugin |
|
|
151
|
-
| **Codex Spark** | gpt-5.3-codex-spark | 초고속 리뷰/린트 (1000+ tok/s) | Codex CLI / Plugin |
|
|
152
|
-
| **Gemini** | gemini-3.1-pro | 리서치, 교차 검증, UI/UX | gemini-cli / API Key |
|
|
136
|
+
Domain-specific skill modules auto-installed based on detected stack.
|
|
153
137
|
|
|
154
|
-
|
|
138
|
+
**Core (15):** Core Capabilities, Parallel Research, Commit Push PR, Git Worktree, Handoff, Priority Todos, Tool Fallback, Context7, Tech Debt, Characterization Test, Agents MD, Claude MD Guide, Exec Plan, Arch Guard, Capability Loop
|
|
155
139
|
|
|
156
|
-
|
|
140
|
+
**Design (7):** Frontend Design, UI/UX Pro Max, Design Teach, Design Audit, Design Critique, Design Polish, Design Normalize
|
|
157
141
|
|
|
158
|
-
|
|
159
|
-
|------|------|
|
|
160
|
-
| **Claude only** | Opus(설계/판단) + Sonnet(리뷰/구현) + Haiku(탐색) |
|
|
161
|
-
| **+Codex** | 구현→5.3-Codex, 리뷰→Spark, 추론→5.4 스위칭 |
|
|
162
|
-
| **+Gemini** | 리서치/리뷰에 Gemini 병렬 추가 |
|
|
163
|
-
| **+Codex +Gemini** | 풀 오케스트레이션 (7개 모델) |
|
|
142
|
+
**Domain (3):** Commerce Patterns, E2E Commerce, Video Production
|
|
164
143
|
|
|
165
|
-
|
|
144
|
+
**PM (3):** Create PRD, Prioritization Frameworks, User Personas
|
|
166
145
|
|
|
167
|
-
|
|
146
|
+
**Event (3):** Event Planning, Event Comms, Event Ops
|
|
168
147
|
|
|
169
|
-
|
|
148
|
+
**Stack-Specific (5):** TypeScript Advanced Types, Vercel React Best Practices, SEO Checklist, Brand Assets, Design Distill
|
|
170
149
|
|
|
171
|
-
|
|
172
|
-
최대 깊이 2단계, 동시 실행 5개, 기본 타임아웃 5분.
|
|
150
|
+
### External Skills (skills.sh)
|
|
173
151
|
|
|
174
|
-
|
|
152
|
+
Install community skills from the [skills.sh](https://skills.sh) ecosystem:
|
|
175
153
|
|
|
176
|
-
|
|
177
|
-
|
|
154
|
+
```bash
|
|
155
|
+
vibe skills add vercel-labs/next-skills
|
|
156
|
+
```
|
|
178
157
|
|
|
179
|
-
|
|
158
|
+
Auto-installed by stack during `vibe init` / `vibe update`:
|
|
180
159
|
|
|
181
|
-
|
|
160
|
+
| Stack | Auto-installed Package |
|
|
161
|
+
|-------|----------------------|
|
|
162
|
+
| `typescript-react` | `vercel-labs/agent-skills` |
|
|
163
|
+
| `typescript-nextjs` | `vercel-labs/agent-skills`, `vercel-labs/next-skills` |
|
|
182
164
|
|
|
183
165
|
---
|
|
184
166
|
|
|
185
|
-
##
|
|
186
|
-
|
|
187
|
-
Agent-Lightning(Microsoft Research)의 핵심 아키텍처 패턴을 적용합니다.
|
|
167
|
+
## Multi-LLM Orchestration
|
|
188
168
|
|
|
189
|
-
|
|
169
|
+
| Provider | Role | Auth |
|
|
170
|
+
|----------|------|------|
|
|
171
|
+
| **Claude** (Opus / Sonnet / Haiku) | SPEC writing, code review, orchestration | Built-in (Claude Code) |
|
|
172
|
+
| **GPT** | Reasoning, architecture, edge-case analysis | Codex CLI / API Key |
|
|
173
|
+
| **Gemini** | Research, cross-validation, UI/UX | gemini-cli / API Key |
|
|
190
174
|
|
|
191
|
-
|
|
175
|
+
### Dynamic Model Routing
|
|
192
176
|
|
|
193
|
-
|
|
194
|
-
import { InMemoryStorage } from '@su-record/vibe/tools';
|
|
195
|
-
const store = new InMemoryStorage(); // DB 없이 테스트
|
|
196
|
-
```
|
|
177
|
+
Auto-switches based on active LLM availability. Defaults to Claude-only operation.
|
|
197
178
|
|
|
198
|
-
|
|
179
|
+
| State | Behavior |
|
|
180
|
+
|-------|----------|
|
|
181
|
+
| **Claude only** | Opus (design/judgment) + Sonnet (review/implementation) + Haiku (exploration) |
|
|
182
|
+
| **+ GPT** | Implementation → GPT, review → GPT, reasoning → GPT |
|
|
183
|
+
| **+ Gemini** | Research/review gets parallel Gemini |
|
|
184
|
+
| **+ GPT + Gemini** | Full orchestration across all models |
|
|
199
185
|
|
|
200
|
-
|
|
201
|
-
|
|
202
|
-
| 스팬 타입 | 용도 |
|
|
203
|
-
|----------|------|
|
|
204
|
-
| `skill_run` | 스킬 실행 |
|
|
205
|
-
| `agent_run` | 에이전트 실행 |
|
|
206
|
-
| `edit` | 파일 수정 |
|
|
207
|
-
| `build` | 빌드/컴파일 |
|
|
208
|
-
| `review` | 코드 리뷰 |
|
|
209
|
-
| `hook` | 훅 실행 |
|
|
210
|
-
| `llm_call` | LLM 호출 |
|
|
211
|
-
| `decision` | 의사결정 |
|
|
212
|
-
|
|
213
|
-
```typescript
|
|
214
|
-
import { createSpan, completeSpan } from '@su-record/vibe/tools';
|
|
215
|
-
|
|
216
|
-
const span = createSpan('agent_run', 'explorer-low', { model: 'haiku' });
|
|
217
|
-
// ... 작업 수행
|
|
218
|
-
const done = completeSpan(span, 'ok', 1500);
|
|
219
|
-
```
|
|
220
|
-
|
|
221
|
-
부모-자식 계층 추적을 지원합니다 (`parent_id`). 모든 데이터는 로컬 JSONL에만 저장됩니다.
|
|
186
|
+
---
|
|
222
187
|
|
|
223
|
-
|
|
188
|
+
## 24 Framework Detection
|
|
224
189
|
|
|
225
|
-
|
|
190
|
+
Auto-detects project stack and applies framework-specific coding rules.
|
|
191
|
+
Supports monorepos (pnpm-workspace, npm workspaces, Lerna, Nx, Turborepo).
|
|
226
192
|
|
|
227
|
-
|
|
228
|
-
|
|
193
|
+
- **TypeScript (12)** — Next.js, React, Angular, Vue, Svelte, Nuxt, NestJS, Node, Electron, Tauri, React Native, Astro
|
|
194
|
+
- **Python (2)** — Django, FastAPI
|
|
195
|
+
- **Java/Kotlin (2)** — Spring Boot, Android
|
|
196
|
+
- **Other** — Rails, Go, Rust, Swift (iOS), Unity (C#), Flutter (Dart), Godot (GDScript)
|
|
229
197
|
|
|
230
|
-
|
|
231
|
-
skills.register('review', () => new ReviewRunner(), { version: '2.0' });
|
|
232
|
-
const runner = skills.resolve('review');
|
|
233
|
-
```
|
|
198
|
+
Also detects: databases (PostgreSQL, MySQL, MongoDB, Redis, Prisma, Drizzle, etc.), state management (Redux, Zustand, Jotai, Pinia, etc.), CI/CD, and hosting platforms.
|
|
234
199
|
|
|
235
200
|
---
|
|
236
201
|
|
|
237
|
-
##
|
|
238
|
-
|
|
239
|
-
SQLite + FTS5 하이브리드 검색으로 세션 간 컨텍스트를 유지합니다.
|
|
240
|
-
|
|
241
|
-
**4가지 엔티티**: Decision, Constraint, Goal, Evidence
|
|
202
|
+
## Orchestrators
|
|
242
203
|
|
|
243
|
-
|
|
244
|
-
최종 점수 = BM25 × 0.4 + 최신성 × 0.3 + 우선순위 × 0.3
|
|
245
|
-
```
|
|
204
|
+
### SwarmOrchestrator
|
|
246
205
|
|
|
247
|
-
|
|
206
|
+
Auto-decomposes tasks with complexity score ≥ 15 into parallel subtasks.
|
|
207
|
+
Max depth 2, concurrent limit 5, default timeout 5 min.
|
|
248
208
|
|
|
249
|
-
|
|
209
|
+
### PhasePipeline
|
|
250
210
|
|
|
251
|
-
|
|
211
|
+
`prepare()` → `execute()` → `cleanup()` lifecycle.
|
|
212
|
+
In ULTRAWORK mode, the next phase's `prepare()` runs in parallel.
|
|
252
213
|
|
|
253
|
-
###
|
|
214
|
+
### BackgroundManager
|
|
254
215
|
|
|
255
|
-
|
|
216
|
+
Per-model/provider concurrency limits. Timeout retry (max 3, exponential backoff). 24-hour TTL auto-cleanup.
|
|
256
217
|
|
|
257
|
-
|
|
218
|
+
---
|
|
258
219
|
|
|
259
|
-
|
|
220
|
+
## Infrastructure
|
|
260
221
|
|
|
261
|
-
|
|
222
|
+
### Session RAG
|
|
262
223
|
|
|
263
|
-
|
|
224
|
+
SQLite + FTS5 hybrid search for cross-session context persistence.
|
|
264
225
|
|
|
265
|
-
|
|
226
|
+
**4 entity types:** Decision, Constraint, Goal, Evidence
|
|
266
227
|
|
|
267
|
-
```
|
|
268
|
-
|
|
228
|
+
```
|
|
229
|
+
Score = BM25 × 0.4 + Recency × 0.3 + Priority × 0.3
|
|
269
230
|
```
|
|
270
231
|
|
|
271
|
-
|
|
272
|
-
|
|
273
|
-
| 스택 | 자동 설치 패키지 |
|
|
274
|
-
|------|-----------------|
|
|
275
|
-
| `typescript-react` | `vercel-labs/agent-skills` |
|
|
276
|
-
| `typescript-nextjs` | `vercel-labs/agent-skills`, `vercel-labs/next-skills` |
|
|
277
|
-
|
|
278
|
-
---
|
|
279
|
-
|
|
280
|
-
## 25개 프레임워크 지원
|
|
232
|
+
On session start, active Goals, critical Constraints, and recent Decisions are auto-injected.
|
|
281
233
|
|
|
282
|
-
|
|
283
|
-
모노레포 지원 (pnpm-workspace, npm workspaces, Lerna, Nx, Turborepo).
|
|
234
|
+
### Structured Telemetry
|
|
284
235
|
|
|
285
|
-
|
|
286
|
-
- **Python (2)** — Django, FastAPI
|
|
287
|
-
- **Java/Kotlin (2)** — Spring Boot, Android
|
|
288
|
-
- **기타** — Rails, Go, Rust, Swift (iOS), Unity (C#), Flutter (Dart), Godot (GDScript)
|
|
236
|
+
8 typed span kinds track all operations:
|
|
289
237
|
|
|
290
|
-
|
|
238
|
+
`skill_run` · `agent_run` · `edit` · `build` · `review` · `hook` · `llm_call` · `decision`
|
|
291
239
|
|
|
292
|
-
|
|
240
|
+
Parent-child hierarchy via `parent_id`. All data stays in local JSONL.
|
|
293
241
|
|
|
294
|
-
|
|
242
|
+
### Evolution System
|
|
295
243
|
|
|
296
|
-
|
|
244
|
+
Self-improving agent/skill/rule generation with benchmarking:
|
|
297
245
|
|
|
298
|
-
|
|
246
|
+
- Usage tracking and insight extraction
|
|
247
|
+
- Skill gap detection
|
|
248
|
+
- Auto-generation with evaluation runners
|
|
249
|
+
- Circuit breaker and rollback safety
|
|
299
250
|
|
|
300
|
-
###
|
|
251
|
+
### Component Registry
|
|
301
252
|
|
|
302
|
-
|
|
253
|
+
Runtime component registration/resolution with metadata:
|
|
303
254
|
|
|
304
|
-
|
|
255
|
+
```typescript
|
|
256
|
+
import { ComponentRegistry } from '@su-record/vibe/tools';
|
|
305
257
|
|
|
306
|
-
|
|
258
|
+
const skills = new ComponentRegistry<SkillRunner>();
|
|
259
|
+
skills.register('review', () => new ReviewRunner(), { version: '2.0' });
|
|
260
|
+
const runner = skills.resolve('review');
|
|
261
|
+
```
|
|
307
262
|
|
|
308
|
-
|
|
263
|
+
---
|
|
309
264
|
|
|
310
|
-
|
|
265
|
+
## Hooks (16 scripts)
|
|
311
266
|
|
|
312
|
-
|
|
267
|
+
| Event | Script | Role |
|
|
268
|
+
|-------|--------|------|
|
|
269
|
+
| SessionStart | `session-start.js` | Restore session context, load memory |
|
|
270
|
+
| PreToolUse | `pre-tool-guard.js` | Block destructive commands, scope protection |
|
|
271
|
+
| PostToolUse | `code-check.js` | Type safety / complexity verification |
|
|
272
|
+
| PostToolUse | `post-edit.js` | Git index update |
|
|
273
|
+
| UserPromptSubmit | `prompt-dispatcher.js` | Command routing |
|
|
274
|
+
| UserPromptSubmit | `keyword-detector.js` | Magic keyword detection |
|
|
275
|
+
| Notification | `context-save.js` | Auto-save at 80/90/95% context |
|
|
313
276
|
|
|
314
|
-
`
|
|
277
|
+
Additional: `llm-orchestrate.js`, `codex-review-gate.js`, `codex-detect.js`, `sentinel-guard.js`, `skill-injector.js`, `evolution-engine.js`, `hud-status.js`, `stop-notify.js`
|
|
315
278
|
|
|
316
279
|
---
|
|
317
280
|
|
|
318
|
-
##
|
|
281
|
+
## Quality Gates
|
|
319
282
|
|
|
320
|
-
|
|
|
321
|
-
|
|
322
|
-
|
|
|
323
|
-
|
|
|
324
|
-
|
|
|
325
|
-
|
|
|
326
|
-
|
|
|
327
|
-
|
|
|
328
|
-
|
|
|
329
|
-
|
|
|
283
|
+
| Guard | Mechanism |
|
|
284
|
+
|-------|-----------|
|
|
285
|
+
| **Type Safety** | Quality Gate — blocks `any`, `@ts-ignore` |
|
|
286
|
+
| **Code Review** | 12 Sonnet agents parallel review + Codex triple cross-validation |
|
|
287
|
+
| **Boundary Check** | API ↔ Frontend type/routing/state consistency verification |
|
|
288
|
+
| **Completeness** | Ralph Loop — iterates until 100% (no scope reduction) |
|
|
289
|
+
| **Convergence** | P1=0 means done; scope narrows on repeated rounds |
|
|
290
|
+
| **Scope Protection** | pre-tool-guard — prevents out-of-scope modifications |
|
|
291
|
+
| **Context Protection** | context-save — auto-saves at 80/90/95% |
|
|
292
|
+
| **Evidence Gate** | No completion claims without evidence |
|
|
330
293
|
|
|
331
|
-
|
|
294
|
+
**Complexity limits:** Function ≤ 50 lines | Nesting ≤ 3 | Parameters ≤ 5 | Cyclomatic complexity ≤ 10
|
|
332
295
|
|
|
333
296
|
---
|
|
334
297
|
|
|
335
|
-
##
|
|
336
|
-
|
|
337
|
-
|
|
|
338
|
-
|
|
339
|
-
|
|
|
340
|
-
|
|
|
341
|
-
|
|
|
342
|
-
|
|
|
343
|
-
|
|
|
344
|
-
|
|
|
345
|
-
|
|
|
346
|
-
|
|
347
|
-
|
|
298
|
+
## Slash Commands
|
|
299
|
+
|
|
300
|
+
| Command | Description |
|
|
301
|
+
|---------|-------------|
|
|
302
|
+
| `/vibe.spec "feature"` | Write SPEC + GPT/Gemini parallel research |
|
|
303
|
+
| `/vibe.spec.review` | SPEC quality review |
|
|
304
|
+
| `/vibe.run "feature"` | Implement from SPEC + parallel code review |
|
|
305
|
+
| `/vibe.verify "feature"` | BDD verification against SPEC |
|
|
306
|
+
| `/vibe.review` | 12-agent parallel code review |
|
|
307
|
+
| `/vibe.trace "feature"` | Requirements traceability matrix |
|
|
308
|
+
| `/vibe.reason "problem"` | Systematic reasoning framework |
|
|
309
|
+
| `/vibe.analyze` | Project analysis |
|
|
310
|
+
| `/vibe.event` | Event automation |
|
|
311
|
+
| `/vibe.utils` | Utilities (E2E, diagrams, UI, session restore) |
|
|
348
312
|
|
|
349
313
|
---
|
|
350
314
|
|
|
351
|
-
##
|
|
352
|
-
|
|
353
|
-
|
|
|
354
|
-
|
|
355
|
-
|
|
|
356
|
-
|
|
|
357
|
-
|
|
|
358
|
-
|
|
|
359
|
-
|
|
|
360
|
-
| `/vibe.trace "기능"` | 요구사항 추적성 매트릭스 |
|
|
361
|
-
| `/vibe.reason "문제"` | 체계적 추론 프레임워크 |
|
|
362
|
-
| `/vibe.analyze` | 프로젝트 분석 |
|
|
363
|
-
| `/vibe.event` | 이벤트 자동화 |
|
|
364
|
-
| `/vibe.utils` | 유틸리티 (E2E, 다이어그램, UI, 세션 복원) |
|
|
315
|
+
## Magic Keywords
|
|
316
|
+
|
|
317
|
+
| Keyword | Effect |
|
|
318
|
+
|---------|--------|
|
|
319
|
+
| `ultrawork` / `ulw` | Parallel processing + phase pipelining + auto-continue + Ralph Loop |
|
|
320
|
+
| `ralph` | Iterate until 100% complete (no scope reduction) |
|
|
321
|
+
| `ralplan` | Iterative planning + persistence |
|
|
322
|
+
| `verify` | Strict verification mode |
|
|
323
|
+
| `quick` | Fast mode, minimal verification |
|
|
365
324
|
|
|
366
325
|
---
|
|
367
326
|
|
|
368
327
|
## CLI
|
|
369
328
|
|
|
370
329
|
```bash
|
|
371
|
-
#
|
|
372
|
-
vibe init [project] #
|
|
373
|
-
vibe update #
|
|
374
|
-
vibe upgrade #
|
|
375
|
-
vibe setup #
|
|
376
|
-
vibe status #
|
|
377
|
-
vibe remove #
|
|
378
|
-
|
|
379
|
-
# LLM
|
|
330
|
+
# Project
|
|
331
|
+
vibe init [project] # Initialize project
|
|
332
|
+
vibe update # Update settings (re-detect stacks)
|
|
333
|
+
vibe upgrade # Upgrade to latest version
|
|
334
|
+
vibe setup # Interactive setup wizard
|
|
335
|
+
vibe status # Show status
|
|
336
|
+
vibe remove # Uninstall
|
|
337
|
+
|
|
338
|
+
# LLM Auth
|
|
380
339
|
vibe gpt auth|key|status|logout
|
|
381
340
|
vibe gemini auth|key|status|logout
|
|
382
341
|
vibe claude key|status|logout
|
|
383
342
|
|
|
384
|
-
#
|
|
385
|
-
vibe skills add <owner/repo> # skills.sh
|
|
343
|
+
# External Skills
|
|
344
|
+
vibe skills add <owner/repo> # Install skills from skills.sh
|
|
386
345
|
|
|
387
|
-
#
|
|
346
|
+
# Channels
|
|
388
347
|
vibe telegram setup|chat|status
|
|
389
348
|
vibe slack setup|channel|status
|
|
390
349
|
|
|
391
|
-
#
|
|
392
|
-
vibe env import [path] # .env → config.json
|
|
350
|
+
# Other
|
|
351
|
+
vibe env import [path] # Migrate .env → config.json
|
|
393
352
|
vibe help / version
|
|
394
353
|
```
|
|
395
354
|
|
|
396
|
-
###
|
|
355
|
+
### Auth Priority
|
|
397
356
|
|
|
398
|
-
|
|
|
399
|
-
|
|
357
|
+
| Provider | Priority |
|
|
358
|
+
|----------|----------|
|
|
400
359
|
| **GPT** | Codex CLI → API Key |
|
|
401
|
-
| **Gemini** | gemini-cli
|
|
360
|
+
| **Gemini** | gemini-cli auto-detect → API Key |
|
|
402
361
|
|
|
403
362
|
---
|
|
404
363
|
|
|
405
|
-
##
|
|
364
|
+
## Subpath Exports
|
|
406
365
|
|
|
407
|
-
|
|
408
|
-
|
|
409
|
-
|
|
410
|
-
|
|
411
|
-
|
|
412
|
-
|
|
413
|
-
|
|
366
|
+
```typescript
|
|
367
|
+
import { MemoryStorage, SessionRAGStore } from '@su-record/vibe/memory';
|
|
368
|
+
import { SwarmOrchestrator, PhasePipeline } from '@su-record/vibe/orchestrator';
|
|
369
|
+
import { findSymbol, validateCodeQuality } from '@su-record/vibe/tools';
|
|
370
|
+
import { InMemoryStorage, ComponentRegistry, createSpan } from '@su-record/vibe/tools';
|
|
371
|
+
```
|
|
372
|
+
|
|
373
|
+
| Subpath | Key Exports |
|
|
374
|
+
|---------|-------------|
|
|
375
|
+
| `@su-record/vibe/memory` | `MemoryStorage`, `IMemoryStorage`, `InMemoryStorage`, `KnowledgeGraph`, `SessionRAGStore` |
|
|
376
|
+
| `@su-record/vibe/orchestrator` | `SwarmOrchestrator`, `PhasePipeline`, `BackgroundManager` |
|
|
377
|
+
| `@su-record/vibe/tools` | `findSymbol`, `validateCodeQuality`, `createSpan`, `ComponentRegistry`, etc. |
|
|
378
|
+
| `@su-record/vibe/tools/memory` | Memory tools |
|
|
379
|
+
| `@su-record/vibe/tools/convention` | Code quality tools |
|
|
380
|
+
| `@su-record/vibe/tools/semantic` | Semantic analysis (symbol search, AST, LSP) |
|
|
381
|
+
| `@su-record/vibe/tools/ui` | UI/UX tools |
|
|
382
|
+
| `@su-record/vibe/tools/interaction` | User interaction tools |
|
|
383
|
+
| `@su-record/vibe/tools/time` | Time utilities |
|
|
414
384
|
|
|
415
385
|
---
|
|
416
386
|
|
|
417
|
-
##
|
|
387
|
+
## Configuration
|
|
418
388
|
|
|
419
|
-
###
|
|
389
|
+
### Global: `~/.vibe/config.json`
|
|
420
390
|
|
|
421
|
-
|
|
391
|
+
Auth, channels, and model settings (file permissions `0o600`).
|
|
422
392
|
|
|
423
393
|
```json
|
|
424
394
|
{
|
|
@@ -434,70 +404,43 @@ vibe help / version
|
|
|
434
404
|
}
|
|
435
405
|
```
|
|
436
406
|
|
|
437
|
-
###
|
|
438
|
-
|
|
439
|
-
프로젝트별 설정 — language, quality, stacks, details, references, installedExternalSkills.
|
|
440
|
-
|
|
441
|
-
---
|
|
442
|
-
|
|
443
|
-
## 모듈 서브패스 Export
|
|
407
|
+
### Project: `.claude/vibe/config.json`
|
|
444
408
|
|
|
445
|
-
|
|
446
|
-
|
|
447
|
-
```typescript
|
|
448
|
-
import { MemoryStorage, SessionRAGStore } from '@su-record/vibe/memory';
|
|
449
|
-
import { SwarmOrchestrator, PhasePipeline } from '@su-record/vibe/orchestrator';
|
|
450
|
-
import { findSymbol, validateCodeQuality } from '@su-record/vibe/tools';
|
|
451
|
-
|
|
452
|
-
// Agent-Lightning 패턴
|
|
453
|
-
import { InMemoryStorage, ComponentRegistry, createSpan } from '@su-record/vibe/tools';
|
|
454
|
-
```
|
|
455
|
-
|
|
456
|
-
| 서브패스 | 주요 export |
|
|
457
|
-
|---------|------------|
|
|
458
|
-
| `@su-record/vibe/memory` | `MemoryStorage`, `IMemoryStorage`, `InMemoryStorage`, `KnowledgeGraph`, `SessionRAGStore` |
|
|
459
|
-
| `@su-record/vibe/orchestrator` | `SwarmOrchestrator`, `PhasePipeline`, `BackgroundManager` |
|
|
460
|
-
| `@su-record/vibe/tools` | `findSymbol`, `validateCodeQuality`, `createSpan`, `ComponentRegistry` 등 |
|
|
461
|
-
| `@su-record/vibe/tools/memory` | 메모리 도구 |
|
|
462
|
-
| `@su-record/vibe/tools/convention` | 코드 품질 도구 |
|
|
463
|
-
| `@su-record/vibe/tools/semantic` | 시맨틱 분석 도구 |
|
|
464
|
-
| `@su-record/vibe/tools/ui` | UI/UX 도구 |
|
|
465
|
-
| `@su-record/vibe/tools/interaction` | 사용자 상호작용 도구 |
|
|
466
|
-
| `@su-record/vibe/tools/time` | 시간 도구 |
|
|
409
|
+
Per-project settings — language, quality, stacks, details, references, installedExternalSkills.
|
|
467
410
|
|
|
468
411
|
---
|
|
469
412
|
|
|
470
|
-
##
|
|
413
|
+
## Project Structure
|
|
471
414
|
|
|
472
415
|
```
|
|
473
416
|
your-project/
|
|
474
417
|
├── .claude/
|
|
475
418
|
│ ├── vibe/
|
|
476
|
-
│ │ ├── config.json #
|
|
477
|
-
│ │ ├── constitution.md #
|
|
478
|
-
│ │ ├── specs/ # SPEC
|
|
479
|
-
│ │ ├── features/ #
|
|
480
|
-
│ │ ├── todos/ # P1/P2/P3
|
|
481
|
-
│ │ └── reports/ #
|
|
482
|
-
│ └── skills/ #
|
|
483
|
-
├── CLAUDE.md #
|
|
484
|
-
├── AGENTS.md # Codex CLI
|
|
419
|
+
│ │ ├── config.json # Project config
|
|
420
|
+
│ │ ├── constitution.md # Project principles
|
|
421
|
+
│ │ ├── specs/ # SPEC documents
|
|
422
|
+
│ │ ├── features/ # Feature tracking
|
|
423
|
+
│ │ ├── todos/ # P1/P2/P3 issues
|
|
424
|
+
│ │ └── reports/ # Review reports
|
|
425
|
+
│ └── skills/ # Local + external skills
|
|
426
|
+
├── CLAUDE.md # Project guide (auto-generated)
|
|
427
|
+
├── AGENTS.md # Codex CLI guide (auto-generated)
|
|
485
428
|
└── ...
|
|
486
429
|
|
|
487
|
-
~/.vibe/config.json #
|
|
488
|
-
~/.vibe/analytics/ #
|
|
489
|
-
│ ├── skill-usage.jsonl
|
|
490
|
-
│ ├── spans.jsonl
|
|
491
|
-
│ └── decisions.jsonl
|
|
430
|
+
~/.vibe/config.json # Global config (auth, channels, models)
|
|
431
|
+
~/.vibe/analytics/ # Telemetry (local JSONL)
|
|
432
|
+
│ ├── skill-usage.jsonl
|
|
433
|
+
│ ├── spans.jsonl
|
|
434
|
+
│ └── decisions.jsonl
|
|
492
435
|
~/.claude/
|
|
493
436
|
├── vibe/
|
|
494
|
-
│ ├── rules/ #
|
|
495
|
-
│ ├── skills/ #
|
|
496
|
-
│ └── ui-ux-data/ # UI/UX CSV
|
|
497
|
-
├── commands/ #
|
|
498
|
-
└── agents/ #
|
|
437
|
+
│ ├── rules/ # Coding rules
|
|
438
|
+
│ ├── skills/ # Global skills
|
|
439
|
+
│ └── ui-ux-data/ # UI/UX CSV datasets (48 files)
|
|
440
|
+
├── commands/ # Slash commands
|
|
441
|
+
└── agents/ # Agent definitions (56)
|
|
499
442
|
~/.codex/
|
|
500
|
-
└── plugins/vibe/ # Codex
|
|
443
|
+
└── plugins/vibe/ # Codex plugin
|
|
501
444
|
├── .codex-plugin/plugin.json
|
|
502
445
|
├── agents/
|
|
503
446
|
├── skills/
|
|
@@ -506,45 +449,44 @@ your-project/
|
|
|
506
449
|
|
|
507
450
|
---
|
|
508
451
|
|
|
509
|
-
##
|
|
452
|
+
## Architecture
|
|
510
453
|
|
|
511
454
|
```mermaid
|
|
512
455
|
flowchart TD
|
|
513
|
-
A["
|
|
456
|
+
A["User Request"] --> B["keyword-detector"]
|
|
514
457
|
B --> C["prompt-dispatcher"]
|
|
515
458
|
C --> D["SmartRouter"]
|
|
516
459
|
|
|
517
460
|
D --> E["LLMCluster"]
|
|
518
|
-
E --> E1["GPT
|
|
519
|
-
E --> E2["
|
|
520
|
-
E --> E3["Gemini (리서치)"]
|
|
461
|
+
E --> E1["GPT (Reasoning)"]
|
|
462
|
+
E --> E2["Gemini (Research)"]
|
|
521
463
|
|
|
522
464
|
D --> F["PhasePipeline"]
|
|
523
465
|
F --> G["SwarmOrchestrator"]
|
|
524
466
|
G --> H["BackgroundManager"]
|
|
525
467
|
H --> I["AgentRegistry"]
|
|
526
468
|
|
|
527
|
-
D --> J["
|
|
528
|
-
J --> J1["12
|
|
529
|
-
J --> J2["
|
|
530
|
-
J1 --> K["
|
|
469
|
+
D --> J["Parallel Review"]
|
|
470
|
+
J --> J1["12 Sonnet Review Agents"]
|
|
471
|
+
J --> J2["Codex Review"]
|
|
472
|
+
J1 --> K["Triple Cross-Validation → P1/P2/P3"]
|
|
531
473
|
J2 --> K
|
|
532
|
-
J --> J3["
|
|
474
|
+
J --> J3["Boundary Verification"]
|
|
533
475
|
|
|
534
476
|
L["Session RAG"] -.-> M["Decision / Constraint / Goal / Evidence"]
|
|
535
|
-
N["VibeSpan"] -.-> O["spans.jsonl (
|
|
536
|
-
P["
|
|
537
|
-
P -.-> R["code-check →
|
|
477
|
+
N["VibeSpan"] -.-> O["spans.jsonl (Local Telemetry)"]
|
|
478
|
+
P["Quality Gate"] -.-> Q["pre-tool-guard → Block"]
|
|
479
|
+
P -.-> R["code-check → Verify"]
|
|
538
480
|
```
|
|
539
481
|
|
|
540
482
|
---
|
|
541
483
|
|
|
542
|
-
##
|
|
484
|
+
## Requirements
|
|
543
485
|
|
|
544
486
|
- **Node.js** >= 18.0.0
|
|
545
|
-
- **Claude Code** (
|
|
546
|
-
- GPT, Gemini (
|
|
487
|
+
- **Claude Code** (required)
|
|
488
|
+
- GPT, Gemini (optional — for multi-LLM features)
|
|
547
489
|
|
|
548
|
-
##
|
|
490
|
+
## License
|
|
549
491
|
|
|
550
492
|
MIT License - Copyright (c) 2025 Su
|