maestro-flow 0.3.46 → 0.3.47

package/.codex/skills/maestro-brainstorm/SKILL.md

@@ -1,451 +1,129 @@
----
-name: maestro-brainstorm
-description: Brainstorm with auto pipeline or single-role analysis
-argument-hint: "[topic] [-y|--yes] [-c|--concurrency N] [--continue] [--count N] [--skip-questions]"
-allowed-tools: spawn_agents_on_csv, Read, Write, Edit, Bash, Glob, Grep, request_user_input
----
-
-<purpose>
-Wave-based multi-role brainstorming using `spawn_agents_on_csv`. Diamond topology: guidance specification generation (Wave 1), parallel role analysis agents (Wave 2), then synthesis + feature-index generation (Wave 3).
-
-**Core workflow**: Parse Topic -> Generate Guidance Spec -> Parallel Role Analysis -> Synthesis + Feature Index
-
-```
-+---------------------------------------------------------------------------+
-|                  BRAINSTORM CSV WAVE WORKFLOW                              |
-+---------------------------------------------------------------------------+
-|                                                                           |
-|  Phase 1: Topic Resolution -> CSV                                         |
-|     +-- Parse topic and flags from arguments                              |
-|     +-- Detect mode (phase / scratch)                                     |
-|     +-- Resolve output directory                                          |
-|     +-- Select roles (interactive or auto, --count N)                     |
-|     +-- Load project specs for architecture-aware analysis                |
-|     +-- Generate tasks.csv with guidance + role + synthesis rows           |
-|     +-- User validates role breakdown (skip if -y)                        |
-|                                                                           |
-|  Phase 2: Wave Execution Engine                                           |
-|     +-- Wave 1: Guidance Specification (single agent)                     |
-|     |   +-- Analyze topic, extract terminology, define boundaries         |
-|     |   +-- Generate guidance-specification.md with RFC 2119 keywords     |
-|     |   +-- Feature decomposition (max 8 features)                        |
-|     |   +-- Discoveries shared via board (terms, non-goals, features)     |
-|     |   +-- Results: guidance_spec content + analysis_file path           |
-|     +-- [CHECKPOINT] User reviews guidance spec (skip if -y)              |
-|     +-- Wave 2: Role Analysis (parallel, 3-9 agents)                      |
-|     |   +-- Each role agent analyzes topic through its lens               |
-|     |   +-- Receives guidance-specification.md via prev_context           |
-|     |   +-- Feature-point organization when feature list available        |
-|     |   +-- Discoveries shared via board (role insights, conflicts)       |
-|     |   +-- Results: analysis_file path per role                          |
-|     +-- [CHECKPOINT] User reviews role analyses (skip if -y)              |
-|     +-- Wave 3: Synthesis + Feature Index (single agent)                  |
-|     |   +-- Cross-role analysis: consensus, conflicts, unique insights    |
-|     |   +-- Generate feature specs or synthesis-specification.md          |
-|     |   +-- Build feature-index.json + synthesis-changelog.md             |
-|     |   +-- Discoveries shared via board (cross-role patterns)            |
-|     +-- discoveries.ndjson shared across all waves (append-only)          |
-|                                                                           |
-|  Phase 3: Results Aggregation                                             |
-|     +-- Export results.csv                                                |
-|     +-- Generate context.md with all findings                             |
-|     +-- Copy artifacts to .brainstorming/ directory                       |
-|     +-- Update phase index.json with brainstorm status                    |
-|     +-- [ROUTING] Offer ui-design when UI features detected               |
-|     +-- Display summary with next steps                                   |
-|                                                                           |
-+---------------------------------------------------------------------------+
-```
-
-</purpose>
-
-<context>
-```bash
-$maestro-brainstorm "Build real-time collaboration platform"
-$maestro-brainstorm -y "Build real-time collaboration platform"
-$maestro-brainstorm -c 6 "Build real-time collaboration platform --count 5"
-$maestro-brainstorm --continue "20260318-brainstorm-collab"
-```
-
-**Flags**:
-- `-y, --yes`: Skip all confirmations (auto mode)
-- `-c, --concurrency N`: Max concurrent agents within each wave (default: 6)
-- `--continue`: Resume existing session
-- `--count N`: Number of roles to select (default 3, max 9)
-- `--skip-questions`: Skip context gathering questions
-
-When `--yes` or `-y`: Auto-confirm role selection, skip interactive questions, use defaults for count and role selection.
-
-**Output Directory**: `.workflow/.csv-wave/{session-id}/`
-**Core Output**: `tasks.csv` (master state) + `results.csv` (final) + `discoveries.ndjson` (shared exploration) + `context.md` (human-readable report) + `.brainstorming/` (role analyses + synthesis artifacts)
-</context>
-
-<csv_schema>
-
-### tasks.csv (Master State)
-
-```csv
-id,title,description,role,topic,guidance_spec,deps,context_from,wave,status,findings,analysis_file,error
-"1","Guidance Specification","Analyze topic and generate guidance-specification.md: extract 5-10 core terms, define non-goals, decompose into features (max 8), apply RFC 2119 keywords. Output: guidance-specification.md with sections for positioning, terminology, non-goals, feature decomposition.","guidance-generator","Build real-time collaboration platform","","","","1","","","",""
-"2","System Architect Analysis","Analyze topic from system-architect perspective: technical architecture, scalability, integration patterns. Must include Data Model (3-5 entities), State Machine, Error Handling, Observability (5+ metrics), Configuration Model. Use RFC 2119 keywords. Reference guidance-specification.md for framework.","system-architect","Build real-time collaboration platform","","1","1","2","","","",""
-"3","UI Designer Analysis","Analyze topic from ui-designer perspective: visual design, layout systems, component hierarchy, design tokens, responsive breakpoints. Reference guidance-specification.md for framework and feature decomposition.","ui-designer","Build real-time collaboration platform","","1","1","2","","","",""
-"4","Product Manager Analysis","Analyze topic from product-manager perspective: product strategy, prioritization, user value proposition, competitive analysis, roadmap alignment. Reference guidance-specification.md for framework.","product-manager","Build real-time collaboration platform","","1","1","2","","","",""
-"5","Synthesis + Feature Index","Cross-role synthesis: analyze all role outputs for consensus, conflicts, unique contributions. Generate feature specs (F-001 through F-00N) or synthesis-specification.md. Build feature-index.json and synthesis-changelog.md. Resolve conflicts with [RESOLVED]/[SUGGESTED]/[UNRESOLVED] tags.","synthesis","Build real-time collaboration platform","","2;3;4","2;3;4","3","","","",""
-```
-
-**Columns**:
-
-| Column | Phase | Description |
-|--------|-------|-------------|
-| `id` | Input | Unique task identifier (string) |
-| `title` | Input | Short task title |
-| `description` | Input | Detailed instructions for this task |
-| `role` | Input | Role identifier: guidance-generator, system-architect, ui-designer, product-manager, ..., synthesis |
-| `topic` | Input | Brainstorm topic text |
-| `guidance_spec` | Input | Guidance-specification.md content reference (populated for wave 2+ from wave 1) |
-| `deps` | Input | Semicolon-separated dependency task IDs |
-| `context_from` | Input | Semicolon-separated task IDs whose findings this task needs |
-| `wave` | Computed | Wave number (1 = guidance, 2 = role analysis, 3 = synthesis) |
-| `status` | Output | `pending` -> `completed` / `failed` / `skipped` |
-| `findings` | Output | Key findings summary (max 500 chars) |
-| `analysis_file` | Output | Path to generated analysis file |
-| `error` | Output | Error message if failed |
-
-### Per-Wave CSV (Temporary)
-
-Each wave generates `wave-{N}.csv` with extra `prev_context` column.
-
-### Output Artifacts
-
-| File | Purpose | Lifecycle |
-|------|---------|-----------|
-| `tasks.csv` | Master state -- all tasks with status/findings | Updated after each wave |
-| `wave-{N}.csv` | Per-wave input (temporary) | Created before wave, deleted after |
-| `results.csv` | Final export of all task results | Created in Phase 3 |
-| `discoveries.ndjson` | Shared exploration board | Append-only, carries across waves |
-| `context.md` | Human-readable brainstorm report | Created in Phase 3 |
-| `.brainstorming/` | Role analyses + synthesis artifacts | Populated in Phase 3 |
-
-### Session Structure
-
-```
-.workflow/.csv-wave/{YYYYMMDD}-brainstorm-{slug}/
-+-- tasks.csv
-+-- results.csv
-+-- discoveries.ndjson
-+-- context.md
-+-- wave-{N}.csv (temporary)
-+-- .brainstorming/
-    +-- guidance-specification.md
-    +-- feature-index.json
-    +-- synthesis-changelog.md
-    +-- feature-specs/
-    |   +-- F-001-{slug}.md
-    |   +-- F-00N-{slug}.md
-    +-- {role}/
-    |   +-- analysis.md
-    |   +-- analysis-cross-cutting.md
-    |   +-- analysis-F-{id}-{slug}.md
-    +-- synthesis-specification.md (fallback mode)
-```
-</csv_schema>
-
-<invariants>
-1. **Start Immediately**: First action is session initialization, then Phase 1
-2. **Wave Order is Sacred**: Never execute wave 2 before wave 1 completes and results are merged
-3. **CSV is Source of Truth**: Master tasks.csv holds all state
-4. **Context Propagation**: prev_context built from master CSV, not from memory
-5. **Guidance First**: Wave 1 (guidance) MUST complete before any role analysis begins
-6. **Discovery Board is Append-Only**: Never clear, modify, or recreate discoveries.ndjson
-7. **Skip on Failure**: If guidance fails, abort. If all roles fail, skip synthesis.
-8. **Cleanup Temp Files**: Remove wave-{N}.csv after results are merged
-9. **DO NOT STOP**: Continuous execution until all waves complete — only pause at [CHECKPOINT] interaction points (skipped with -y)
-10. **9 Valid Roles Only**: data-architect, product-manager, product-owner, scrum-master, subject-matter-expert, system-architect, test-strategist, ui-designer, ux-expert
-</invariants>
-
-<execution>
-
-### Session Initialization
-
-```
-Parse from $ARGUMENTS:
-  AUTO_YES       ← --yes | -y
-  continueMode   ← --continue
-  maxConcurrency ← --concurrency | -c N  (default: 6)
-  roleCount      ← --count N  (default: 3, max: 9)
-  skipQuestions   ← --skip-questions
-  topicArg       ← remaining text after flag removal
-
-Derive:
-  slug           ← topicArg kebab-cased, max 40 chars
-  dateStr        ← UTC+8 YYYYMMDD
-  sessionId      ← "{dateStr}-brainstorm-{slug}"
-  sessionFolder  ← ".workflow/.csv-wave/{sessionId}"
-
-mkdir -p {sessionFolder}/.brainstorming
-```
-
-### Phase 1: Topic Resolution -> CSV
-
-**Objective**: Parse topic, select roles, generate tasks.csv.
-
-**Decomposition Rules**:
-
-1. **Mode detection**: Number = phase mode (resolve via state.json artifact registry to `.workflow/scratch/{YYYYMMDD}-{type}-{slug}/`), text = scratch mode
-2. **Project specs loading**: Run `maestro spec load --category arch` for architecture-aware analysis context and discoverable knowhow tools
-3. **Role selection**:
-
-| Condition | Action |
-|-----------|--------|
-| `--yes` flag | Auto-select top `roleCount` roles based on topic relevance |
-| Interactive | `request_user_input` with recommended roles + rationale |
-
-4. **Valid roles** (9 total):
-
-| Role ID | Focus Area |
-|---------|------------|
-| `data-architect` | Data models, storage strategies, data flow |
-| `product-manager` | Product strategy, roadmap, prioritization |
-| `product-owner` | Backlog management, user stories, acceptance criteria |
-| `scrum-master` | Process facilitation, impediment removal |
-| `subject-matter-expert` | Domain knowledge, business rules, compliance |
-| `system-architect` | Technical architecture, scalability, integration |
-| `test-strategist` | Test strategy, quality assurance |
-| `ui-designer` | Visual design, mockups, design systems |
-| `ux-expert` | User research, information architecture, journey |
-
-5. **CSV generation**: 1 guidance row (wave 1) + N role rows (wave 2) + 1 synthesis row (wave 3).
-
-**Wave computation**: 3-wave diamond -- guidance = wave 1, all role tasks = wave 2, synthesis = wave 3.
-
-**User validation**: Display task breakdown (skip if AUTO_YES).
-
-### Phase 2: Wave Execution Engine
-
-**Objective**: Execute brainstorm pipeline wave-by-wave via spawn_agents_on_csv.
-
-#### Wave 1: Guidance Specification (Single Agent)
-
-1. Extract wave 1 pending rows from master `tasks.csv` into `wave-1.csv` (no prev_context needed)
-2. Execute:
-
-```javascript
-spawn_agents_on_csv({
-  csv_path: `${sessionFolder}/wave-1.csv`,
-  id_column: "id",
-  instruction: buildGuidanceInstruction(sessionFolder, topicArg),
-  max_concurrency: 1, max_runtime_seconds: 3600,
-  output_csv_path: `${sessionFolder}/wave-1-results.csv`,
-  output_schema: { id, status: [completed|failed], findings, analysis_file, error }
-})
-```
-
-3. Merge results into master `tasks.csv`, delete `wave-1.csv`
-4. Read generated `guidance-specification.md` for wave 2 context propagation
-5. **Interaction Checkpoint** (skip if AUTO_YES):
-   - Display guidance-specification summary: terminology count, feature list, non-goals
-   - `request_user_input`:
-     ```json
-     { "questions": [{ "id": "guidance_review", "header": "Guidance", "question": "Guidance specification generated. How to proceed?", "options": [
-       { "label": "Proceed (Recommended)", "description": "Continue to parallel role analysis with current guidance." },
-       { "label": "Revise", "description": "Provide feedback to directly edit guidance-specification.md." },
-       { "label": "Abort", "description": "Stop pipeline, output partial results (guidance only)." }
-     ]}] }
-     ```
-     - **Proceed** → continue to Wave 2
-     - **Revise** → user provides feedback, directly edit guidance-specification.md to incorporate changes
-     - **Abort** → stop pipeline, output partial results (guidance only)
-
-**Guidance agent responsibilities**:
-- Analyze topic, extract 5-10 core domain terms
-- Define non-goals and scope boundaries
-- Decompose into features (max 8): F-{3-digit} ID, kebab-case slug, description, related roles, priority
-- Apply RFC 2119 keywords (MUST, SHOULD, MAY, MUST NOT, SHOULD NOT)
-- Write `guidance-specification.md` to `.brainstorming/`
-
-#### Wave 2: Role Analysis (Parallel, 3-9 Agents)
-
-1. Extract wave 2 pending rows from master `tasks.csv`
-2. Build `prev_context` from wave 1 findings + `guidance-specification.md` content
-3. Write `wave-2.csv` with `prev_context` column
-4. Execute:
-
-```javascript
-spawn_agents_on_csv({
-  csv_path: `${sessionFolder}/wave-2.csv`,
-  id_column: "id",
-  instruction: buildRoleAnalysisInstruction(sessionFolder),  // agent: ~/.codex/agents/conceptual-planning-agent.toml
-  max_concurrency: maxConcurrency, max_runtime_seconds: 3600,
-  output_csv_path: `${sessionFolder}/wave-2-results.csv`,
-  output_schema: { id, status: [completed|failed], findings, analysis_file, error }
-})
-```
-
-5. Merge results into master `tasks.csv`, delete `wave-2.csv`
-6. **Interaction Checkpoint** (skip if AUTO_YES):
-   - Display per-role findings summary (role name + key insight, 1-2 lines each)
-   - Show success/failure counts: "{N} roles completed, {M} failed"
-   - `request_user_input`:
-     ```json
-     { "questions": [{ "id": "role_review", "header": "Roles", "question": "Role analyses complete. How to proceed?", "options": [
-       { "label": "Proceed (Recommended)", "description": "Continue to cross-role synthesis." },
-       { "label": "Add Roles", "description": "Select additional roles and run only the new ones." },
-       { "label": "Revise", "description": "Pick role(s) to directly edit analysis files with feedback." }
-     ]}] }
-     ```
-     - **Proceed** → continue to Wave 3 (synthesis)
-     - **Add Roles** → select additional roles, generate new wave-2 rows, execute only new roles
-     - **Revise** → user picks specific role(s), directly edit its analysis files to incorporate feedback
-
-**Role agent responsibilities**:
-- Read guidance-specification.md for framework context
-- Analyze topic through role-specific lens
-- Feature-point organization when feature list available:
-  - `analysis.md` -- Role overview INDEX only (< 1500 words)
-  - `analysis-cross-cutting.md` -- Cross-feature decisions (< 2000 words)
-  - `analysis-F-{id}-{slug}.md` -- Per-feature analysis (< 2000 words each)
-- Fallback organization (no feature list):
-  - `analysis.md` -- Main analysis (< 3000 words)
-- system-architect MUST include: Data Model, State Machine, Error Handling, Observability, Configuration Model
-- Write analysis files to `.brainstorming/{role}/`
-
-#### Wave 3: Synthesis + Feature Index (Single Agent)
-
-1. If all wave 2 tasks failed, skip synthesis entirely
-2. Extract wave 3 pending rows, build `prev_context` from wave 2 findings
-3. Write `wave-3.csv`, execute `spawn_agents_on_csv` for synthesis agent (same output_schema pattern)
-4. Merge results into master `tasks.csv`, delete `wave-3.csv`
-
-**Synthesis agent responsibilities**:
-- Cross-role analysis: consensus, conflicts, unique contributions
-- Conflict resolution with [RESOLVED]/[SUGGESTED]/[UNRESOLVED] tags
-- Feature mode: generate `feature-specs/F-{id}-{slug}.md` per feature (7 sections, 1500-2500 words)
-- Fallback mode: generate `synthesis-specification.md`
-- Build `feature-index.json` and `synthesis-changelog.md`
-- Four-Layer Aggregation: Direct Reference, Structured Extraction, Conflict Distillation, Cross-Feature Annotation
-
-### Phase 3: Results Aggregation
-
-**Objective**: Generate final results and human-readable report.
-
-1. Export final `tasks.csv` as `results.csv`
-2. Generate `context.md`:
-
-```markdown
-# Brainstorm Report -- {topic}
-
-## Summary
-- Topic: {topic}
-- Roles analyzed: {role_count}
-- Features decomposed: {feature_count}
-- Conflicts resolved: {resolved_count}
-
-## Guidance Specification
-{summary of guidance-specification.md}
-
-## Role Analysis Results
-### {role_name}
-{findings summary}
-**Analysis file**: {analysis_file path}
-
-## Synthesis
-{synthesis summary}
-- Consensus areas: {list}
-- Resolved conflicts: {list}
-- Unresolved items: {list}
-
-## Feature Index
-{feature-index.json summary}
-
-## Next Steps
-- Skill: maestro-ui-design {topic} -- Generate UI prototypes (when UI features detected)
-- Skill: maestro-analyze -- Evaluate feasibility + lock decisions
-- Skill: maestro-plan -- Plan directly if scope is clear
-- Skill: maestro-roadmap --mode full -- Generate full spec package from brainstorm
-```
-
-4. **Brainstorm confidence scoring**:
-
-   Dimensions (5): role_coverage, cross_role_consistency, feature_completeness, spec_quality, design_feasibility. Factors (weights): analysis_depth(.30), evidence_strength(.25), coverage_breadth(.20), user_validation(.15, 0 if --yes), consistency(.10). Append confidence summary to `synthesis-changelog.md`.
-
-   **Conflict-based quality gate**: >3 `[UNRESOLVED]` conflicts → warn before artifact registration.
-
-5. Copy artifacts to output `.brainstorming/` directory (phase mode or scratch mode target)
-6. Update phase `index.json` with brainstorm status (if phase mode)
-7. **Next-Step Routing** (skip if AUTO_YES — default to first applicable):
-   - Detect UI features: scan feature-index.json for UI/frontend-related features (keywords: ui, interface, page, component, dashboard, form, layout)
-   - `request_user_input` (include UI Design option only when UI features detected):
-     ```json
-     { "questions": [{ "id": "next_step", "header": "Next Step", "question": "Brainstorm complete. What next?", "options": [
-       { "label": "UI Design (Recommended)", "description": "Generate UI prototypes from brainstorm features." },
-       { "label": "Analyze", "description": "Evaluate feasibility and lock architectural decisions." },
-       { "label": "Plan", "description": "Create execution plan directly if scope is clear." }
-     ]}] }
-     ```
-     - **UI Design** → invoke `maestro-ui-design` with topic + brainstorm session context
-     - **Analyze** → invoke `maestro-analyze` for feasibility evaluation
-     - **Plan** → invoke `maestro-plan` directly if scope is clear
-     - When no UI features detected, replace UI Design with Roadmap: `{ "label": "Roadmap (Recommended)", "description": "Generate full spec package from brainstorm." }`
-
-### Shared Discovery Board Protocol
-
-#### Standard Discovery Types
-
-| Type | Dedup Key | Data Schema | Description |
-|------|-----------|-------------|-------------|
-| `code_pattern` | `data.name` | `{name, file, description}` | Reusable code pattern found |
-| `integration_point` | `data.file` | `{file, description, exports[]}` | Module connection point |
-| `convention` | singleton | `{naming, imports, formatting}` | Project code conventions |
-| `blocker` | `data.issue` | `{issue, severity, impact}` | Blocking issue found |
-| `tech_stack` | singleton | `{framework, language, tools[]}` | Technology stack info |
-
-#### Domain Discovery Types
-
-| Type | Dedup Key | Data Schema | Description |
-|------|-----------|-------------|-------------|
-| `terminology` | `data.term` | `{term, definition, aliases[], category}` | Domain term extracted |
-| `non_goal` | `data.title` | `{title, rationale}` | Scope exclusion identified |
-| `feature_candidate` | `data.id` | `{id, slug, description, roles[], priority}` | Feature decomposed from topic |
-| `role_insight` | `data.role+data.topic` | `{role, topic, insight, confidence}` | Role-specific finding |
-| `cross_role_conflict` | `data.area` | `{area, roles[], positions[], resolution}` | Cross-role disagreement |
-
-#### Protocol
-
-Read `discoveries.ndjson` before analysis. Append-only: dedup by type+key before writing, never modify/delete.
-
-```bash
-echo '{"ts":"<ISO>","worker":"{id}","type":"terminology","data":{"term":"CRDT","definition":"Conflict-free Replicated Data Type","aliases":["conflict-free"],"category":"technical"}}' >> {session_folder}/discoveries.ndjson
-```
-</execution>
-
-<error_codes>
-
-| Error | Resolution |
-|-------|------------|
-| Topic argument missing | Abort with error: "Topic text required" |
-| Invalid role name in selection | Filter to valid roles, warn user |
-| Role count exceeds 9 | Cap at 9 with warning |
-| Guidance agent timeout | Mark as failed, abort pipeline (wave 2 depends on guidance) |
-| Role agent timeout | Mark as failed, continue with remaining roles |
-| All role agents failed | Skip synthesis (wave 3), report partial results |
-| Synthesis agent failed | Use wave 2 results directly, no feature index |
-| CSV parse error | Validate format, show line number |
-| discoveries.ndjson corrupt | Ignore malformed lines |
-| Continue mode: no session found | List available sessions |
-| Phase directory not found | Abort with error: "Phase {N} not found" |
-</error_codes>
-
-<success_criteria>
-- [ ] Session folder created with valid tasks.csv
-- [ ] All 3 waves executed in order
-- [ ] guidance-specification.md produced in .brainstorming/
-- [ ] Role analysis files produced per selected role
-- [ ] Synthesis artifacts produced (feature specs or synthesis-specification.md)
-- [ ] feature-index.json and synthesis-changelog.md produced
-- [ ] context.md produced with full brainstorm report
-- [ ] Artifacts copied to target .brainstorming/ directory
-- [ ] discoveries.ndjson append-only throughout
-- [ ] Confidence scored per role and after cross-role synthesis
-- [ ] Conflict-based quality gate evaluated (> 3 unresolved = warning)
-- [ ] Confidence summary appended to synthesis-changelog.md
-</success_criteria>
+---
+name: maestro-brainstorm
+description: Brainstorm with auto pipeline or single-role analysis
+argument-hint: "[topic] [-y|--yes] [-c|--concurrency N] [--continue] [--count N] [--skip-questions]"
+allowed-tools: spawn_agents_on_csv, Read, Write, Edit, Bash, Glob, Grep, request_user_input
+---
+
+<purpose>
+Wave-based multi-role brainstorming via `spawn_agents_on_csv`. Diamond topology:
+Wave 1 (guidance spec) → Wave 2 (parallel role analysis, 3-9 agents) → Wave 3 (synthesis + feature index).
+</purpose>
+
+<context>
+$ARGUMENTS — topic text and optional flags.
+
+**Flags**: `-y` (auto), `-c N` (concurrency, default 6), `--continue` (resume), `--count N` (roles, default 3 max 9), `--skip-questions`
+
+**9 valid roles**: data-architect, product-manager, product-owner, scrum-master, subject-matter-expert, system-architect, test-strategist, ui-designer, ux-expert
+
+**Session**: `.workflow/.csv-wave/{YYYYMMDD}-brainstorm-{slug}/`
+**Output**: tasks.csv, results.csv, discoveries.ndjson, context.md, `.brainstorming/` (guidance-specification.md, feature-index.json, synthesis-changelog.md, feature-specs/, {role}/analysis*.md)
+</context>
+
+<csv_schema>
+```csv
+id,title,description,role,topic,guidance_spec,deps,context_from,wave,status,findings,analysis_file,error
+"1","Guidance Spec","...","guidance-generator","<topic>","","","","1","","","",""
+"2","System Architect","...","system-architect","<topic>","","1","1","2","","","",""
+"3","UI Designer","...","ui-designer","<topic>","","1","1","2","","","",""
+"4","Synthesis","...","synthesis","<topic>","","2;3","2;3","3","","","",""
+```
+Wave 1: 1 guidance row. Wave 2: N role rows (parallel). Wave 3: 1 synthesis row.
+</csv_schema>
+
+<invariants>
+1. **Wave order sacred**: Guidance (W1) MUST complete before roles (W2)
+2. **CSV source of truth**: Master tasks.csv holds all state
+3. **Discovery board append-only**: Never modify/delete discoveries.ndjson
+4. **Skip on failure**: Guidance fails → abort. All roles fail → skip synthesis.
+5. **9 valid roles only**: data-architect, product-manager, product-owner, scrum-master, subject-matter-expert, system-architect, test-strategist, ui-designer, ux-expert
+6. **DO NOT STOP**: Continuous until all waves complete; only pause at [CHECKPOINT] (skipped with -y)
+</invariants>
+
+<state_machine>
+
+<states>
+S_PARSE      — 解析 topic、flags、mode                    PERSIST: —
+S_ROLES      — 选择 roles（-y auto / interactive）        PERSIST: —
+S_CSV_GEN    — 生成 tasks.csv                              PERSIST: tasks.csv
+S_WAVE_1     — Guidance Spec (single agent)                PERSIST: guidance-specification.md
+S_CHECK_1    — Checkpoint: 用户审阅 guidance（-y 跳过）    PERSIST: —
+S_WAVE_2     — Role Analysis (parallel spawn)              PERSIST: role analyses
+S_CHECK_2    — Checkpoint: 用户审阅 roles（-y 跳过）       PERSIST: —
+S_WAVE_3     — Synthesis + Feature Index (single agent)    PERSIST: synthesis artifacts
+S_AGGREGATE  — 生成报告、注册 artifact                     PERSIST: context.md + results.csv
+</states>
+
+<transitions>
+S_PARSE → S_ROLES      DO: parse args, detect mode (phase/scratch), load specs
+S_ROLES → S_CSV_GEN    DO: select roles (-y: auto top N; interactive: request_user_input)
+S_CSV_GEN → S_WAVE_1   DO: generate tasks.csv (1 guidance + N roles + 1 synthesis)
+
+S_WAVE_1 → S_CHECK_1   WHEN: completed    DO: spawn wave-1, merge results, read guidance-spec
+S_WAVE_1 → END         WHEN: failed       DO: abort pipeline
+
+S_CHECK_1 → S_WAVE_2   WHEN: -y OR user "Proceed"
+S_CHECK_1 → S_CHECK_1  WHEN: user "Revise"    DO: edit guidance-spec, re-display
+S_CHECK_1 → END        WHEN: user "Abort"
+
+S_WAVE_2 → S_CHECK_2   DO: spawn wave-2 (parallel), merge results
+S_WAVE_2 → S_WAVE_3    WHEN: all failed       DO: skip synthesis
+
+S_CHECK_2 → S_WAVE_3   WHEN: -y OR user "Proceed"
+S_CHECK_2 → S_WAVE_2   WHEN: user "Add Roles"  DO: add new role rows, spawn only new
+
+S_WAVE_3 → S_AGGREGATE DO: spawn wave-3, merge results
+
+S_AGGREGATE → END      DO: A_AGGREGATE
+</transitions>
+
+<actions>
+
+### Wave agent responsibilities
+
+**Guidance (W1)**: Analyze topic → extract 5-10 terms → define non-goals → decompose features (max 8, F-{3digit}) → RFC 2119 keywords → write guidance-specification.md
+
+**Role (W2)**: Read guidance-spec → analyze through role lens → feature-point organization (analysis.md index + analysis-cross-cutting.md + per-feature analysis-F-{id}.md) or fallback single analysis.md. system-architect MUST include: Data Model, State Machine, Error Handling, Observability, Config Model.
+
+**Synthesis (W3)**: Cross-role consensus/conflicts/unique → conflict tags [RESOLVED]/[SUGGESTED]/[UNRESOLVED] → feature-specs or synthesis-specification.md → feature-index.json + synthesis-changelog.md. Four-layer: Direct Reference → Structured Extraction → Conflict Distillation → Cross-Feature Annotation.
+
+### A_AGGREGATE
+
+1. Export results.csv
+2. Generate context.md (summary, guidance, per-role findings, synthesis, feature index, next steps)
+3. Confidence scoring: 5 dimensions (role_coverage, cross_role_consistency, feature_completeness, spec_quality, design_feasibility). Quality gate: >3 UNRESOLVED → warn.
+4. Copy artifacts to target .brainstorming/
+5. Next-step routing: UI features detected → maestro-ui-design; else → maestro-analyze / maestro-plan / maestro-roadmap
+
+</actions>
+</state_machine>
+
+<discovery_board>
+| Type | Dedup Key | Data |
+|------|-----------|------|
+| terminology | term | {term, definition, aliases[], category} |
+| non_goal | title | {title, rationale} |
+| feature_candidate | id | {id, slug, description, roles[], priority} |
+| role_insight | role+topic | {role, topic, insight, confidence} |
+| cross_role_conflict | area | {area, roles[], positions[], resolution} |
+
+Protocol: read before analysis, append-only, dedup by type+key.
+</discovery_board>
+
+<error_codes>
+| Condition | Recovery |
+|-----------|----------|
+| Guidance agent failed | Abort pipeline (W2 depends on guidance) |
+| All role agents failed | Skip synthesis, report partial |
+| Synthesis failed | Use W2 results directly |
+| Role count > 9 | Cap at 9 with warning |
+</error_codes>
+
+<success_criteria>
+- [ ] 3 waves executed: guidance → parallel roles → synthesis
+- [ ] guidance-specification.md + role analyses + synthesis artifacts produced
+- [ ] feature-index.json + context.md generated
+- [ ] discoveries.ndjson append-only throughout
+- [ ] Confidence scored, conflict quality gate evaluated
+</success_criteria>