haki-skills 0.2.3 → 0.2.4

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.
@@ -0,0 +1,810 @@
1
+ ---
2
+ name: swarm-dev-team
3
+ description: Spawn a full software development team as parallel subagents, each with specialized roles and haki skills. Supports both greenfield and brownfield projects. Produces structured reports for tracking and review.
4
+ ---
5
+
6
+ # Swarm Dev Team
7
+
8
+ Orchestrate a complete software development team using Claude Code subagents. Each agent assumes a specialized role, leverages specific haki skills/workflows, and produces a standardized report.
9
+
10
+ **Works with:** Existing haki workflows (`/haki:*`), haki-tools CLI, and all `.agent/skills/`.
11
+
12
+ ---
13
+
14
+ ## When to Use
15
+
16
+ - Starting a new feature or project that spans multiple concerns (DB, backend, frontend, testing, security, docs, DevOps)
17
+ - Applying structured team development to an existing codebase (brownfield)
18
+ - When you need traceable reports per role for review and auditing
19
+
20
+ ## Prerequisites
21
+
22
+ - Haki installed in the project (`npx haki-skills --for claude`)
23
+ - `.haki/PROJECT.md` and `.haki/ROADMAP.md` exist (run `/haki:new-project` first)
24
+ - For brownfield: `.haki/codebase/` exists (run `/haki:map-codebase` first)
25
+
26
+ ---
27
+
28
+ ## Team Architecture
29
+
30
+ ```
31
+ ┌──────────────────────────────────────────────────────────┐
32
+ │ 🎯 TECH LEAD (You / Orchestrator) │
33
+ │ /haki:next → Discovery → Plan → Dispatch → Review │
34
+ └────────────────────────────┬─────────────────────────────┘
35
+
36
+ ┌────────────────────┼─────────────────────┐
37
+ │ │ │
38
+ ▼ ▼ ▼
39
+ ┌──────────────┐ ┌──────────────┐ ┌───────────────┐
40
+ │ 📐 Architect │ │ 🗄️ DB Eng. │ │ 🔬 Researcher │
41
+ │ (Phase 1) │ │ (Phase 2) │ │ (Phase 1) │
42
+ └──────┬───────┘ └──────┬───────┘ └───────────────┘
43
+ │ │
44
+ ▼ ▼
45
+ ┌──────────────┐ ┌──────────────┐ ← Parallel
46
+ │ ⚙️ Backend │ │ 🎨 Frontend │
47
+ │ (Phase 3) │ │ (Phase 3) │
48
+ └──────┬───────┘ └──────┬───────┘
49
+ │ │
50
+ ▼ ▼
51
+ ┌──────────────┐ ┌──────────────┐ ← Parallel
52
+ │ 🧪 QA / Test │ │ 🔒 Security │
53
+ │ (Phase 4) │ │ (Phase 4) │
54
+ └──────┬───────┘ └──────┬───────┘
55
+ │ │
56
+ ▼ ▼
57
+ ┌──────────────┐ ┌──────────────┐ ← Parallel
58
+ │ 📝 Tech Docs │ │ 🚀 DevOps │
59
+ │ (Phase 5) │ │ (Phase 5) │
60
+ └──────────────┘ └──────────────┘
61
+
62
+
63
+ ┌──────────────────────────────────┐
64
+ │ 📊 DASHBOARD (auto-generated) │
65
+ └──────────────────────────────────┘
66
+ ```
67
+
68
+ ---
69
+
70
+ ## Report Directory Structure
71
+
72
+ All agent reports are stored under `.haki/reports/`. This directory is created automatically.
73
+
74
+ ```
75
+ .haki/
76
+ ├── reports/
77
+ │ ├── 00-discovery.md ← Brownfield only: codebase analysis
78
+ │ ├── 01-architect.md ← Architecture decisions & diagrams
79
+ │ ├── 02-researcher.md ← Tech stack research (Context7)
80
+ │ ├── 03-db-engineer.md ← Schema design & migrations
81
+ │ ├── 04-backend-dev.md ← Backend implementation report
82
+ │ ├── 05-frontend-dev.md ← Frontend implementation report
83
+ │ ├── 06-qa-engineer.md ← Test results & coverage
84
+ │ ├── 07-security-engineer.md ← Security audit findings
85
+ │ ├── 08-tech-writer.md ← Documentation status
86
+ │ ├── 09-devops-engineer.md ← CI/CD & deployment report
87
+ │ └── DASHBOARD.md ← Aggregated team dashboard
88
+ ├── PROJECT.md
89
+ ├── ROADMAP.md
90
+ ├── tasks/
91
+ ├── research/
92
+ └── codebase/ ← From /haki:map-codebase
93
+ ```
94
+
95
+ ---
96
+
97
+ ## Report Template (MANDATORY for every agent)
98
+
99
+ Every subagent MUST write its report to `.haki/reports/NN-role.md` using this exact structure:
100
+
101
+ ```markdown
102
+ # [ICON] [ROLE NAME] Report
103
+
104
+ **Agent:** [Role name]
105
+ **Phase:** [Phase number]
106
+ **Status:** 🟢 COMPLETED | 🟡 PARTIAL | 🔴 BLOCKED
107
+ **Started:** [ISO 8601 timestamp]
108
+ **Completed:** [ISO 8601 timestamp]
109
+
110
+ ---
111
+
112
+ ## Input
113
+
114
+ - **Received from:** [Previous agent/phase]
115
+ - **Files read:** [List of files consumed]
116
+ - **Haki context:** [PROJECT.md, ROADMAP.md, task files, codebase maps used]
117
+
118
+ ---
119
+
120
+ ## Objectives
121
+
122
+ - [x] Objective 1 — completed
123
+ - [x] Objective 2 — completed
124
+ - [ ] Objective 3 — skipped (reason)
125
+
126
+ ---
127
+
128
+ ## Tasks Executed
129
+
130
+ ### Task 1: [Name]
131
+
132
+ - **Status:** ✅ Done | ⚠️ Partial | ❌ Failed | ⏭️ Skipped
133
+ - **Haki skill used:** [skill name from .agent/skills/]
134
+ - **Haki workflow used:** [workflow name, if any]
135
+ - **Files changed:**
136
+
137
+ | File | Action | Description |
138
+ |------|--------|-------------|
139
+ | `path/to/file` | Created | Brief description |
140
+ | `path/to/file` | Modified | Brief description |
141
+
142
+ - **Details:** [What was done]
143
+ - **Issues found:** [Any problems discovered]
144
+
145
+ ### Task 2: [Name]
146
+ ...
147
+
148
+ ---
149
+
150
+ ## Metrics
151
+
152
+ | Metric | Before | After | Target |
153
+ |--------|--------|-------|--------|
154
+ | [metric] | [val] | [val] | [val] |
155
+
156
+ ---
157
+
158
+ ## Issues & Risks
159
+
160
+ | # | Severity | Description | Recommendation | Status |
161
+ |---|----------|-------------|----------------|--------|
162
+ | 1 | 🔴 Critical | ... | ... | Open |
163
+ | 2 | 🟡 Medium | ... | ... | Resolved |
164
+ | 3 | 🟢 Low | ... | ... | Open |
165
+
166
+ ---
167
+
168
+ ## Output
169
+
170
+ - **Total files created:** [N]
171
+ - **Total files modified:** [N]
172
+ - **Handoff notes:** [What the next agent needs to know]
173
+
174
+ ---
175
+
176
+ ## Recommendations
177
+
178
+ 1. [Actionable recommendation]
179
+ 2. [Actionable recommendation]
180
+ ```
181
+
182
+ ---
183
+
184
+ ## Phase Execution
185
+
186
+ ### Phase 0: Discovery (Brownfield Only)
187
+
188
+ **Trigger:** Project has existing source code.
189
+
190
+ **Action:** Run `/haki:map-codebase` which spawns 4 parallel mapper subagents:
191
+
192
+ | Mapper | Output | Content |
193
+ |--------|--------|---------|
194
+ | Stack | `.haki/codebase/STACK.md` | Languages, frameworks, build tools, deps |
195
+ | Architecture | `.haki/codebase/ARCHITECTURE.md` | System patterns, data flow, API design |
196
+ | Conventions | `.haki/codebase/CONVENTIONS.md` | Code style, naming, testing patterns |
197
+ | Structure | `.haki/codebase/STRUCTURE.md` | Directory tree, entry points, configs |
198
+
199
+ After mapping, also scan for issues:
200
+
201
+ ```bash
202
+ # Find TODOs, FIXMEs, HACKs
203
+ grep -rn "TODO\|FIXME\|HACK\|XXX\|DEPRECATED" --include="*.ts" --include="*.tsx" --include="*.py" --include="*.go" . | head -50
204
+
205
+ # Count test files
206
+ find . -name "*.test.*" -o -name "*.spec.*" -o -name "*_test.*" | wc -l
207
+
208
+ # Check git history
209
+ git log --oneline -20 2>/dev/null
210
+ ```
211
+
212
+ Write discovery summary to `.haki/reports/00-discovery.md` using the report template.
213
+
214
+ **Brownfield golden rules:**
215
+ - Do NOT refactor working code just for style preferences
216
+ - Do NOT change tech stack unless there is a critical reason
217
+ - Prioritize CONSISTENCY with existing codebase conventions
218
+ - Every change must have a rollback path
219
+ - Read `.haki/codebase/CONVENTIONS.md` before writing any code
220
+
221
+ ---
222
+
223
+ ### Phase 1: Architecture & Research (Parallel)
224
+
225
+ Spawn two subagents in parallel:
226
+
227
+ #### 📐 Subagent: System Architect
228
+
229
+ **Haki skills to read first:**
230
+ - `.agent/skills/writing-plans/SKILL.md`
231
+ - `.agent/skills/brainstorming/SKILL.md`
232
+
233
+ **Haki context to load:**
234
+ - `.haki/PROJECT.md` — project vision and constraints
235
+ - `.haki/ROADMAP.md` — task breakdown
236
+ - `.haki/codebase/ARCHITECTURE.md` — existing architecture (brownfield)
237
+ - `.haki/codebase/CONVENTIONS.md` — existing conventions (brownfield)
238
+
239
+ **Instructions:**
240
+
241
+ 1. **Architecture design:**
242
+ - Choose architectural pattern (monolith / modular monolith / microservices)
243
+ - Define layer architecture (presentation → application → domain → infrastructure)
244
+ - Draw system diagrams (C4 model: context, container, component)
245
+ - For brownfield: map as-is → define to-be → plan migration path
246
+
247
+ 2. **API contract design:**
248
+ - Define all API endpoints (OpenAPI spec or markdown table)
249
+ - Request/response schemas with TypeScript types
250
+ - Error response format (standardized)
251
+ - Pagination, filtering, sorting conventions
252
+
253
+ 3. **Coding conventions (greenfield only):**
254
+ - Project folder structure
255
+ - Naming conventions
256
+ - Error handling strategy
257
+ - Logging strategy
258
+ - For brownfield: document EXISTING conventions, do not invent new ones
259
+
260
+ 4. **Create base config files (greenfield only):**
261
+ - `tsconfig.json` / `pyproject.toml` / equivalent
262
+ - Linter config (ESLint/Biome/Ruff)
263
+ - `.env.example`
264
+
265
+ **Output:** Write report to `.haki/reports/01-architect.md`
266
+ **Also save:** `docs/ARCHITECTURE.md`, `docs/api-contract.md`
267
+
268
+ #### 🔬 Subagent: Tech Researcher
269
+
270
+ **Haki skills to read first:**
271
+ - `.agent/skills/context7-research/SKILL.md`
272
+
273
+ **Haki workflow:** Follow `/haki:research` protocol exactly.
274
+
275
+ **Instructions:**
276
+
277
+ 1. For every library in the chosen tech stack, use Context7 MCP:
278
+ - Resolve library ID
279
+ - Query docs for latest version, install command, config format
280
+ - Check peer dependency compatibility
281
+
282
+ 2. Research integration patterns between chosen libraries
283
+
284
+ 3. Save verified stack to `.haki/research/STACK.md`
285
+
286
+ **Output:** Write report to `.haki/reports/02-researcher.md`
287
+
288
+ ---
289
+
290
+ ### Phase 2: Database Engineering
291
+
292
+ #### 🗄️ Subagent: Database Engineer
293
+
294
+ **Haki skills to read first:**
295
+ - `.agent/skills/test-driven-development/SKILL.md`
296
+ - `.agent/skills/subagent-driven-development/SKILL.md`
297
+
298
+ **Haki context to load:**
299
+ - `.haki/reports/01-architect.md` — architecture decisions
300
+ - `.haki/reports/02-researcher.md` — verified library versions
301
+ - `.haki/codebase/STACK.md` — existing DB setup (brownfield)
302
+
303
+ **Instructions:**
304
+
305
+ 1. **Data modeling:**
306
+ - Analyze entities and relationships from ROADMAP requirements
307
+ - Design ERD (Entity Relationship Diagram)
308
+ - Normalize to 3NF; denormalize intentionally for performance
309
+ - Define index strategy
310
+
311
+ 2. **Schema implementation:**
312
+ - Create migration files in dependency order
313
+ - Define constraints: PK, FK, UNIQUE, CHECK, NOT NULL
314
+ - Add audit fields: `created_at`, `updated_at`, `created_by`, `updated_by`
315
+ - Soft delete via `deleted_at` timestamp
316
+ - For brownfield: create NEW migrations only, NEVER edit existing ones
317
+
318
+ 3. **Seed data:**
319
+ - Development seed files
320
+ - Test fixtures
321
+
322
+ 4. **Performance:**
323
+ - Composite indexes for frequent query patterns
324
+ - Identify potential N+1 query patterns
325
+
326
+ **TDD approach:** Write migration → verify it runs → write seed → verify data integrity.
327
+
328
+ **Output:** Write report to `.haki/reports/03-db-engineer.md`
329
+
330
+ ---
331
+
332
+ ### Phase 3: Implementation (Parallel)
333
+
334
+ Spawn two subagents in parallel. Both MUST follow TDD.
335
+
336
+ #### ⚙️ Subagent: Backend Developer
337
+
338
+ **Haki skills to read first:**
339
+ - `.agent/skills/subagent-driven-development/SKILL.md` — follow the implementer prompt
340
+ - `.agent/skills/test-driven-development/SKILL.md` — TDD cycle is mandatory
341
+ - `.agent/skills/systematic-debugging/SKILL.md` — for when tests fail unexpectedly
342
+ - `.agent/skills/verification-before-completion/SKILL.md` — verify before marking done
343
+
344
+ **Haki workflow integration:**
345
+ - For each task from ROADMAP, the plan in `.haki/tasks/[task-id].md` defines the steps
346
+ - Follow the TDD cycle from the task plan: write test → RED → implement → GREEN → commit
347
+ - Update task file's "Implementation Details" and "Execution Results" sections after completion
348
+
349
+ **Haki context to load:**
350
+ - `.haki/reports/01-architect.md` — API contracts, architecture
351
+ - `.haki/reports/03-db-engineer.md` — DB schema, models
352
+ - `.haki/tasks/[task-id].md` — task plan with TDD steps
353
+ - `.haki/codebase/CONVENTIONS.md` — existing code style (brownfield)
354
+
355
+ **Instructions:**
356
+
357
+ 1. **Project setup (greenfield):**
358
+ - Initialize project with verified tech stack from researcher report
359
+ - Install dependencies using exact versions from `.haki/research/STACK.md`
360
+ - Configure project structure per architect report
361
+
362
+ 2. **Core infrastructure:**
363
+ - Database connection & ORM setup
364
+ - Authentication & authorization middleware (JWT/OAuth2)
365
+ - Request validation middleware (Zod/Joi/Pydantic)
366
+ - Global error handler with standardized error format
367
+ - Structured logging middleware
368
+ - Rate limiting, CORS, health check endpoint
369
+
370
+ 3. **Business logic (per task):**
371
+ - Follow Clean Architecture layers:
372
+ - Controllers/Handlers → Services/Use Cases → Repositories → Entities
373
+ - Implement each API endpoint per contract from architect
374
+ - Transaction management for complex operations
375
+ - TDD: every function has a test BEFORE implementation
376
+
377
+ 4. **For brownfield:**
378
+ - Match existing code patterns exactly
379
+ - Add new endpoints following existing router conventions
380
+ - Do NOT restructure existing code
381
+
382
+ **Quality gates:**
383
+ - All tests pass
384
+ - No linting errors
385
+ - No `any` types (TypeScript)
386
+ - Every public function has JSDoc/TSDoc
387
+
388
+ **Output:** Write report to `.haki/reports/04-backend-dev.md`
389
+ **Also update:** `.haki/tasks/[task-id].md` with Implementation Details + Execution Results
390
+
391
+ #### 🎨 Subagent: Frontend Developer
392
+
393
+ **Haki skills to read first:**
394
+ - `.agent/skills/subagent-driven-development/SKILL.md`
395
+ - `.agent/skills/test-driven-development/SKILL.md`
396
+ - Read UI skill based on config: `node .agent/bin/haki-tools.cjs config get ui_design_skill --raw`
397
+ - If `ui-ux-pro-max` → read `.agent/skills/ui-ux-pro-max/SKILL.md`
398
+ - If `taste-skill` → read `.agent/skills/taste-skill/SKILL.md` + variant skill
399
+ - `.agent/skills/verification-before-completion/SKILL.md`
400
+
401
+ **Haki context to load:**
402
+ - `.haki/reports/01-architect.md` — API contracts, component architecture
403
+ - `.haki/reports/02-researcher.md` — verified frontend library versions
404
+ - `.haki/tasks/[task-id].md` — task plan with TDD steps
405
+ - `.haki/codebase/CONVENTIONS.md` — existing UI patterns (brownfield)
406
+
407
+ **Instructions:**
408
+
409
+ 1. **Project setup (greenfield):**
410
+ - Initialize frontend with verified versions
411
+ - Configure build tools, Tailwind/CSS, path aliases
412
+ - Apply chosen UI design skill for design tokens and component style
413
+
414
+ 2. **Design system & components:**
415
+ - Create design tokens (colors, spacing, typography, shadows)
416
+ - Build base components: Button, Input, Select, Modal, Toast, Table, Card
417
+ - Each component must have: TypeScript props, variants, ARIA attributes, responsive design
418
+ - Loading/error/empty states for every data-fetching component
419
+
420
+ 3. **Pages & features (per task):**
421
+ - Implement routing structure
422
+ - API integration using TanStack Query / SWR
423
+ - Form handling with validation (React Hook Form + Zod)
424
+ - Optimistic updates for better UX
425
+
426
+ 4. **Performance & UX:**
427
+ - Code splitting & lazy loading
428
+ - Skeleton loaders, error boundaries
429
+ - Dark mode support (if specified)
430
+
431
+ 5. **For brownfield:**
432
+ - Match existing component patterns
433
+ - Use existing design system/tokens
434
+ - Do NOT introduce new UI libraries unless approved
435
+
436
+ **Output:** Write report to `.haki/reports/05-frontend-dev.md`
437
+ **Also update:** `.haki/tasks/[task-id].md` with Implementation Details + Execution Results
438
+
439
+ ---
440
+
441
+ ### Phase 4: Quality & Security (Parallel)
442
+
443
+ #### 🧪 Subagent: QA / Test Engineer
444
+
445
+ **Haki skills to read first:**
446
+ - `.agent/skills/test-driven-development/SKILL.md`
447
+ - `.agent/skills/api-testing/SKILL.md`
448
+ - `.agent/skills/playwright-automation/SKILL.md`
449
+ - `.agent/skills/playwright-intent-to-spec/SKILL.md`
450
+ - `.agent/skills/systematic-debugging/SKILL.md`
451
+ - `.agent/skills/verification-before-completion/SKILL.md`
452
+
453
+ **Haki workflows to use:**
454
+ - `/haki:api-test init` → set up API test infrastructure
455
+ - `/haki:api-test generate [resource]` → generate API tests per resource
456
+ - `/haki:e2e init` → set up Playwright infrastructure
457
+ - `/haki:e2e generate [feature]` → generate E2E tests per feature
458
+
459
+ **Instructions:**
460
+
461
+ 1. **Review existing tests:**
462
+ - Scan all `*.test.*` and `*.spec.*` files
463
+ - Identify untested code paths using coverage reports
464
+ - Check for testing anti-patterns (read `.agent/skills/test-driven-development/testing-anti-patterns.md`)
465
+
466
+ 2. **Unit test gaps:**
467
+ - Add tests for edge cases: null inputs, boundary values, error scenarios
468
+ - Target: ≥ 80% code coverage
469
+
470
+ 3. **API integration tests:**
471
+ - Use `/haki:api-test` workflow to generate tests for each API resource
472
+ - Test: success (200/201), validation error (400), not found (404), unauthorized (401)
473
+
474
+ 4. **E2E tests:**
475
+ - Use `/haki:e2e` workflow to set up Playwright
476
+ - Generate specs for critical user journeys using `/haki:e2e-gen`
477
+ - Critical paths: registration → login → core feature → logout
478
+
479
+ 5. **Performance baseline:**
480
+ - API response time benchmarks
481
+ - Frontend Lighthouse audit script
482
+
483
+ **Metrics to report:**
484
+
485
+ | Metric | Value | Target |
486
+ |--------|-------|--------|
487
+ | Unit test coverage | X% | ≥ 80% |
488
+ | API tests passing | X/Y | 100% |
489
+ | E2E tests passing | X/Y | 100% |
490
+ | Lint errors | N | 0 |
491
+
492
+ **Output:** Write report to `.haki/reports/06-qa-engineer.md`
493
+
494
+ #### 🔒 Subagent: Security Engineer
495
+
496
+ **Haki skills to read first:**
497
+ - `.agent/skills/systematic-debugging/SKILL.md` — root cause analysis methodology
498
+ - `.agent/skills/verification-before-completion/SKILL.md`
499
+
500
+ **Instructions:**
501
+
502
+ 1. **OWASP Top 10 audit:**
503
+ - Injection (SQL, NoSQL, OS Command)
504
+ - Broken authentication
505
+ - Sensitive data exposure
506
+ - Broken access control
507
+ - XSS (Cross-Site Scripting)
508
+ - Security misconfiguration
509
+ - Using components with known vulnerabilities
510
+
511
+ 2. **Authentication hardening:**
512
+ - Password hashing review (bcrypt/argon2 with proper rounds)
513
+ - JWT security (expiry, refresh rotation, blacklisting)
514
+ - RBAC/ABAC implementation review
515
+ - Brute force protection
516
+
517
+ 3. **Data protection:**
518
+ - Input sanitization audit
519
+ - Output encoding
520
+ - File upload security
521
+ - PII handling
522
+
523
+ 4. **Infrastructure security:**
524
+ - Security headers (CSP, HSTS, X-Frame-Options)
525
+ - CORS policy review
526
+ - Dependency vulnerability scan:
527
+ ```bash
528
+ npm audit 2>/dev/null || bun audit 2>/dev/null || pip audit 2>/dev/null
529
+ ```
530
+ - Secrets management review (.env files, hardcoded secrets)
531
+
532
+ 5. **Fix critical issues immediately.** Log all findings with severity.
533
+
534
+ **Output:** Write report to `.haki/reports/07-security-engineer.md`
535
+
536
+ ---
537
+
538
+ ### Phase 5: Documentation & DevOps (Parallel)
539
+
540
+ #### 📝 Subagent: Technical Writer
541
+
542
+ **Haki skills to read first:**
543
+ - `.agent/skills/user-docs-generator/SKILL.md`
544
+ - `.agent/skills/output-skill/SKILL.md` — full output enforcement
545
+
546
+ **Haki workflow to use:**
547
+ - `/haki:docs --all` — generate user guides for all modules
548
+
549
+ **Instructions:**
550
+
551
+ 1. **README.md:**
552
+ - Project overview, features, tech stack
553
+ - Quick start guide (< 5 minutes to run)
554
+ - Contributing guidelines
555
+
556
+ 2. **Developer documentation:**
557
+ - `docs/GETTING_STARTED.md` — detailed setup
558
+ - `docs/ARCHITECTURE.md` — from architect report
559
+ - `docs/API.md` — API reference from contracts
560
+ - `docs/DATABASE.md` — schema documentation
561
+ - `docs/DEPLOYMENT.md` — deployment guide
562
+
563
+ 3. **Code documentation:**
564
+ - JSDoc/TSDoc for all public functions
565
+ - Inline comments for complex logic
566
+
567
+ 4. **User documentation:**
568
+ - Run `/haki:docs --all` to generate user guides with screenshots
569
+ - Output in `.haki/generated/docs/user-guides/`
570
+
571
+ 5. **ADRs (Architecture Decision Records):**
572
+ - Extract decisions from `.haki/reports/01-architect.md`
573
+ - Write to `docs/adr/NNN-decision-title.md`
574
+
575
+ **Output:** Write report to `.haki/reports/08-tech-writer.md`
576
+
577
+ #### 🚀 Subagent: DevOps Engineer
578
+
579
+ **Haki skills to read first:**
580
+ - `.agent/skills/verification-before-completion/SKILL.md`
581
+
582
+ **Instructions:**
583
+
584
+ 1. **Containerization:**
585
+ - Dockerfile (multi-stage build, non-root user, minimal base image)
586
+ - docker-compose.yml (app + DB + Redis + reverse proxy)
587
+ - `.dockerignore`
588
+
589
+ 2. **CI/CD pipeline:**
590
+ - `.github/workflows/ci.yml`:
591
+ - Lint & format check
592
+ - Unit tests + coverage
593
+ - Integration tests (API tests from QA)
594
+ - Security scan (Trivy/Snyk)
595
+ - Build Docker image
596
+ - E2E tests
597
+ - Deploy (staging/production)
598
+
599
+ 3. **Environment management:**
600
+ - `.env.example` with all variables documented
601
+ - Environment-specific configs (dev/staging/prod)
602
+
603
+ 4. **Monitoring & observability:**
604
+ - Health check endpoints
605
+ - Structured JSON logging config
606
+ - Error tracking setup (Sentry config)
607
+
608
+ 5. **Developer scripts:**
609
+ - `scripts/setup.sh` — one-command dev environment setup
610
+ - `scripts/seed.sh` — database seeding
611
+ - `Makefile` with common commands
612
+
613
+ **Output:** Write report to `.haki/reports/09-devops-engineer.md`
614
+
615
+ ---
616
+
617
+ ## Dashboard Generation
618
+
619
+ After ALL agents complete, generate `.haki/reports/DASHBOARD.md`:
620
+
621
+ ```markdown
622
+ # 📊 Swarm Team Dashboard
623
+
624
+ **Project:** [from .haki/PROJECT.md]
625
+ **Generated:** [ISO 8601 timestamp]
626
+ **Overall Status:** 🟢 All Clear | 🟡 Has Warnings | 🔴 Has Blockers
627
+
628
+ ---
629
+
630
+ ## Team Summary
631
+
632
+ | # | Agent | Status | Tasks | Files Changed | Issues | Duration |
633
+ |---|-------|--------|-------|---------------|--------|----------|
634
+ | 0 | 🔍 Discovery | 🟢 | 4/4 | 4 created | 0 | ~2min |
635
+ | 1 | 📐 Architect | 🟢 | 3/3 | 5 created | 0 | ~3min |
636
+ | 2 | 🔬 Researcher | 🟢 | 1/1 | 1 created | 0 | ~2min |
637
+ | 3 | 🗄️ DB Engineer | 🟢 | 4/4 | 8 created | 1 🟡 | ~3min |
638
+ | 4 | ⚙️ Backend | 🟢 | N/N | X files | 0 | ~Xmin |
639
+ | 5 | 🎨 Frontend | 🟢 | N/N | X files | 0 | ~Xmin |
640
+ | 6 | 🧪 QA | 🟡 | N/N | X files | 2 🟡 | ~Xmin |
641
+ | 7 | 🔒 Security | 🟢 | N/N | X files | 1 🟢 | ~Xmin |
642
+ | 8 | 📝 Docs | 🟢 | N/N | X files | 0 | ~Xmin |
643
+ | 9 | 🚀 DevOps | 🟢 | N/N | X files | 0 | ~Xmin |
644
+
645
+ ---
646
+
647
+ ## Quality Metrics
648
+
649
+ | Metric | Value | Target | Status |
650
+ |--------|-------|--------|--------|
651
+ | Test coverage | X% | ≥ 80% | ✅/❌ |
652
+ | API tests | X/Y pass | 100% | ✅/❌ |
653
+ | E2E tests | X/Y pass | 100% | ✅/❌ |
654
+ | Security critical | N | 0 | ✅/❌ |
655
+ | Lint errors | N | 0 | ✅/❌ |
656
+ | Docs coverage | X% | 100% | ✅/❌ |
657
+
658
+ ---
659
+
660
+ ## Open Issues (aggregated from all reports)
661
+
662
+ | # | Agent | Severity | Description | Recommendation |
663
+ |---|-------|----------|-------------|----------------|
664
+ | ... | ... | ... | ... | ... |
665
+
666
+ ---
667
+
668
+ ## Files Changed (total)
669
+
670
+ - **Created:** [N] files
671
+ - **Modified:** [N] files
672
+ - **Deleted:** [N] files
673
+
674
+ ---
675
+
676
+ ## Recommendations (top priority)
677
+
678
+ 1. [From security report — highest severity first]
679
+ 2. [From QA report]
680
+ 3. [From architect report]
681
+
682
+ ---
683
+
684
+ ## Next Steps
685
+
686
+ - [ ] Review all reports in `.haki/reports/`
687
+ - [ ] Address open issues by severity
688
+ - [ ] Run `/haki:next` to continue with remaining ROADMAP tasks
689
+ ```
690
+
691
+ ---
692
+
693
+ ## Invocation
694
+
695
+ ### Quick Start (Recommended)
696
+
697
+ For a new project:
698
+ ```
699
+ /haki:new-project
700
+ # ... answer questions, get PROJECT.md + ROADMAP.md ...
701
+ # Then:
702
+ /swarm-team
703
+ ```
704
+
705
+ For an existing project:
706
+ ```
707
+ /haki:map-codebase
708
+ # ... wait for codebase analysis ...
709
+ /haki:new-project
710
+ # ... answer questions ...
711
+ /swarm-team
712
+ ```
713
+
714
+ ### Full Orchestration Prompt
715
+
716
+ ```
717
+ Read `.agent/skills/swarm-dev-team/SKILL.md`.
718
+
719
+ This is a [NEW | EXISTING] project.
720
+ Project: [NAME]
721
+ Description: [DESCRIPTION]
722
+ Tech stack: [TECH STACK or "decide based on requirements"]
723
+
724
+ Execute the full swarm team workflow:
725
+ 1. Phase 0: Discovery (if existing project — use /haki:map-codebase)
726
+ 2. Phase 1: Architect + Researcher (parallel subagents)
727
+ 3. Phase 2: DB Engineer
728
+ 4. Phase 3: Backend + Frontend (parallel subagents)
729
+ 5. Phase 4: QA + Security (parallel subagents)
730
+ 6. Phase 5: Docs + DevOps (parallel subagents)
731
+ 7. Generate DASHBOARD.md
732
+
733
+ Every agent MUST:
734
+ - Read its assigned haki skills BEFORE starting work
735
+ - Load context from previous agent reports
736
+ - Follow TDD (test first) for all implementation
737
+ - Write its report to .haki/reports/NN-role.md using the report template
738
+ - Update .haki/tasks/[task-id].md with Implementation Details + Execution Results
739
+
740
+ For brownfield: read .haki/codebase/CONVENTIONS.md and match existing patterns.
741
+ ```
742
+
743
+ ### Step-by-Step (Manual Control)
744
+
745
+ If you prefer to run each phase manually with review between phases:
746
+
747
+ ```bash
748
+ # Phase 0 (brownfield only)
749
+ /haki:map-codebase
750
+
751
+ # Phase 1 — parallel
752
+ # Subagent 1: "You are SYSTEM ARCHITECT. Read .agent/skills/swarm-dev-team/SKILL.md Phase 1 Architect section. Load .haki/PROJECT.md and .haki/ROADMAP.md. Write report to .haki/reports/01-architect.md"
753
+ # Subagent 2: "You are TECH RESEARCHER. Run /haki:research for the full tech stack. Write report to .haki/reports/02-researcher.md"
754
+
755
+ # Review Phase 1 reports before continuing...
756
+
757
+ # Phase 2
758
+ # "You are DB ENGINEER. Read .agent/skills/swarm-dev-team/SKILL.md Phase 2. Load reports 01 and 02. Write report to .haki/reports/03-db-engineer.md"
759
+
760
+ # Phase 3 — parallel
761
+ # Subagent 1: "You are BACKEND DEVELOPER. Read .agent/skills/swarm-dev-team/SKILL.md Phase 3 Backend section..."
762
+ # Subagent 2: "You are FRONTEND DEVELOPER. Read .agent/skills/swarm-dev-team/SKILL.md Phase 3 Frontend section..."
763
+
764
+ # Phase 4 — parallel
765
+ # Subagent 1: "You are QA ENGINEER. Use /haki:api-test and /haki:e2e workflows..."
766
+ # Subagent 2: "You are SECURITY ENGINEER..."
767
+
768
+ # Phase 5 — parallel
769
+ # Subagent 1: "You are TECHNICAL WRITER. Use /haki:docs --all..."
770
+ # Subagent 2: "You are DEVOPS ENGINEER..."
771
+
772
+ # Generate dashboard
773
+ # "Read all .haki/reports/*.md and generate .haki/reports/DASHBOARD.md using the dashboard template from the skill."
774
+ ```
775
+
776
+ ---
777
+
778
+ ## Integration with Haki State Machine
779
+
780
+ This skill integrates with haki's existing state machine:
781
+
782
+ | Haki State | Swarm Action |
783
+ |------------|-------------|
784
+ | No `.haki/` | Run `/haki:new-project` first |
785
+ | Has PROJECT.md, no ROADMAP | Run `/haki:new-project` to completion |
786
+ | Has ROADMAP, tasks ⏳ Pending | Run `/haki:discuss` per task, then swarm |
787
+ | Has ROADMAP, tasks 💬 Discussed | Run `/haki:plan` per task, then swarm |
788
+ | Has ROADMAP, tasks 📋 Planned | **Ready for swarm** — run `/swarm-team` |
789
+ | All tasks ✅ Complete | Run `/haki:new-milestone` for next phase |
790
+
791
+ The swarm skill does NOT replace haki workflows — it **orchestrates them at scale** by assigning specialized roles to subagents and adding a reporting layer.
792
+
793
+ ---
794
+
795
+ ## Quality Gates
796
+
797
+ Each phase must pass before the next begins:
798
+
799
+ | Phase | Gate | Criteria |
800
+ |-------|------|----------|
801
+ | 0 Discovery | ✅ Codebase Mapped | All 4 codebase docs exist in `.haki/codebase/` |
802
+ | 1 Architecture | ✅ Design Complete | API contracts defined, architecture documented |
803
+ | 1 Research | ✅ Stack Verified | All libraries verified via Context7 |
804
+ | 2 Database | ✅ Schema Ready | Migrations run successfully, seeds load |
805
+ | 3 Backend | ✅ APIs Working | All endpoints match contract, tests pass |
806
+ | 3 Frontend | ✅ UI Functional | Components render, responsive, accessible |
807
+ | 4 QA | ✅ Coverage Met | ≥ 80% unit, API tests pass, E2E pass |
808
+ | 4 Security | ✅ No Criticals | Zero critical/high severity findings |
809
+ | 5 Docs | ✅ Docs Complete | README + API docs + user guides exist |
810
+ | 5 DevOps | ✅ Pipeline Green | CI/CD runs end-to-end successfully |
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "haki-skills",
3
- "version": "0.2.3",
3
+ "version": "0.2.4",
4
4
  "description": "AI workflow system — project initialization, task planning, and execution with TDD-first approach",
5
5
  "bin": {
6
6
  "haki-skills": "./bin/install.js",