ccw

## Auto Mode When `-y` or `--yes`: Skip clarification and confirmation prompts. Pass `-y` through to each step's skill invocation. # CCW ## Usage ```bash $ccw "implement user authentication with JWT" $ccw -y "refactor the payment module" $ccw --continue $ccw --dry-run "add rate limiting to API endpoints" $ccw --chain feature "add dark mode toggle" ``` **Flags**: - `-y, --yes` — Auto mode: skip all prompts; propagate `-y` to each skill - `--continue` — Resume latest paused session from last incomplete wave - `--dry-run` — Display planned chain without executing - `--chain <name>` — Force a specific chain (skips intent classification) **Session state**: `.workflow/.ccw-coordinate/{session-id}/` **Core Output**: `tasks.csv` (master) + `wave-{N}-results.csv` (per wave) + `context.md` (report) --- ## Overview Wave-based pipeline coordinator. The coordinator loop builds one wave CSV at a time, calls `spawn_agents_on_csv`, then performs **coordinator-side artifact analysis** before assembling the next wave. Barrier skills produce artifacts (plan.json, analysis results, etc.) that the coordinator reads to dynamically resolve args for subsequent steps. ``` Intent → Structured Extract → Resolve Chain → [Wave Loop]: (action×object×style) (chainMap) ┌─────────────────────────────────────────────────┐ │ 1. Identify next wave (1 or N parallel steps) │ │ 2. Build wave-{N}.csv with skill_call per row │ │ 3. spawn_agents_on_csv(wave-{N}.csv) │ │ 4. Read wave-{N}-results.csv │ │ 5. If barrier skill: analyze artifacts, │ │ update context for subsequent steps │ │ 6. Merge into master tasks.csv │ └─────────────────────────────────────────────────┘ → Report ``` --- ## Barrier Skills Skills that produce artifacts requiring **coordinator-side analysis** before the next wave can be assembled. After a barrier skill completes, the coordinator reads its output and updates the execution context. | Skill | Artifacts to Read | Context Updates | |-------|------------------|-----------------| | `analyze-with-file` | `.workflow/.analysis/ANL-*/conclusions.json` | `analysis_dir`, `gaps`, `phase` | | `brainstorm-with-file` | `.workflow/.brainstorm/*/` | `brainstorm_dir`, `features` | | `workflow-plan` | `.workflow/active/WFS-*/workflow-session.json` | `plan_dir`, `task_count` | | `workflow-lite-planex` | `.workflow/.lite-plan/*/plan.json` | `plan_dir`, `task_count` | | `spec-generator` | `.workflow/.spec/*/` | `spec_session_id` | | `roadmap-with-file` | `.workflow/.roadmap/*/roadmap.md` | `roadmap_dir` | | `workflow-tdd-plan` | `.workflow/.tdd-plan/*/` | `tdd_plan_dir` | | `issue-discover` | `.workflow/.issues/*/` | `issue_dir`, `issue_count` | | `debug-with-file` | `.workflow/.debug/*/` | `debug_dir`, `findings` | **Non-barrier skills** (can be grouped into multi-task waves): `workflow-execute`, `workflow-test-fix-cycle`, `review-cycle`, `clean`, `investigate`, `security-audit`, `ship`, `parallel-dev-cycle`, `brainstorm`, all `team-*` skills --- ## Phase 1: Structured Intent Extraction Extract a structured intent tuple using LLM semantic understanding, then route deterministically via an action × object matrix. **Extract structured intent from user input:** ```json { "action": "<from action enum>", "object": "<from object enum>", "scope": "<module/file/area or null>", "style": "<from style enum>", "urgency": "<low | normal | high>" } ``` **Action enum**: | action | Semantic meaning | |--------|-----------------| | `create` | Build something new — feature, project, component, spec | | `fix` | Repair something broken — fix bug, resolve error, patch | | `analyze` | Understand deeply — analyze, investigate, discuss, explore concept | | `plan` | Design approach — plan, break down, roadmap, decompose | | `execute` | Implement planned work — execute, implement, develop | | `explore` | Open-ended discovery — brainstorm, ideate, creative thinking | | `debug` | Diagnose failures — debug, diagnose, troubleshoot | | `test` | Run or create tests — test, generate test, TDD | | `review` | Evaluate code quality — review, code review | | `refactor` | Restructure code — refactor, clean up, tech debt | | `convert` | Bridge between workflows — convert brainstorm to issue | **Object enum**: | object | Meaning | |--------|---------| | `feature` | New functionality or enhancement | | `bug` | Defect, error, broken behavior | | `issue` | Issue-tracker item for batch/structured management | | `code` | Source code in general | | `test` | Tests, test suite, test coverage | | `spec` | Specification, PRD, product requirements | | `doc` | Documentation | | `ui` | User interface, design, component | | `performance` | Performance characteristics | | `security` | Security concerns | | `architecture` | System architecture, design decisions | | `project` | Entire project (greenfield) | | `team` | Team-based execution | **Style enum**: | style | Meaning | |-------|---------| | `quick` | Fast, lightweight, minimal ceremony | | `documented` | With file artifacts, discussion docs | | `collaborative` | Multi-agent, multi-perspective | | `structured` | Formal planning, spec-driven, phased | | `iterative` | Cycle-based, self-iterating with reflection | | `tdd` | Test-driven development | | `default` | No specific style preference | --- ## Chain Map Routing via `detectTaskType(intent)` → chain name → skill list. ### Task Type Detection (action × object × style matrix) ```javascript function detectTaskType(intent) { const { action, object, style, urgency } = intent; // Urgency override if (urgency === 'high' && (action === 'fix' || object === 'bug')) return 'bugfix-hotfix'; // Style-first routing if (style === 'tdd') return 'tdd'; if (style === 'collaborative' && action === 'plan') return 'collaborative-plan'; if (style === 'collaborative' && action === 'analyze') return 'analyze-wave'; if (style === 'collaborative' && action !== 'plan') return 'multi-cli'; if (style === 'iterative' && object === 'test') return 'integration-test'; if (style === 'iterative' && action === 'refactor') return 'refactor'; // Action × Object matrix const matrix = { 'create': { 'project': 'greenfield', 'feature': 'feature', 'spec': 'spec-driven', 'test': 'test-gen', 'doc': 'documentation', 'ui': 'ui-design', 'issue': 'issue-batch', '_default': 'feature' }, 'fix': { 'bug': 'bugfix', 'test': 'test-fix', 'issue': 'issue-batch', 'code': 'bugfix', 'security': 'bugfix', '_default': 'bugfix' }, 'analyze':{ 'architecture': 'analyze-file', 'code': 'analyze-file', 'bug': 'debug-file', 'security': 'security', '_default': 'analyze-file' }, 'explore':{ 'feature': 'brainstorm', 'architecture': 'brainstorm', 'issue': 'issue-batch', '_default': 'exploration' }, 'plan': { 'feature': 'feature', 'project': 'greenfield', 'issue': 'issue-transition', '_default': 'feature' }, 'execute':{ 'issue': 'issue-transition', '_default': 'feature' }, 'debug': { 'bug': style === 'documented' ? 'debug-file' : 'debug', '_default': style === 'documented' ? 'debug-file' : 'debug' }, 'test': { 'test': 'test-fix', 'code': 'test-gen', 'feature': 'integration-test', '_default': 'test-gen' }, 'review': { '_default': 'review' }, 'refactor':{ '_default': 'refactor' }, 'convert':{ 'issue': 'brainstorm-to-issue', '_default': 'issue-transition' }, }; // Special compound detections if (action === 'plan' && style === 'structured' && /roadmap/.test(rawInput)) return 'roadmap'; if (/csv.?wave|wave.?pipeline|并行波|波次执行/.test(rawInput)) return 'analyze-wave'; if (object === 'team') return 'team-planex'; if (/ship|release|publish/.test(rawInput)) return 'ship'; const actionMap = matrix[action]; if (!actionMap) return 'feature'; return actionMap[object] || actionMap['_default'] || 'feature'; } ``` ### Available Skills Inventory Skills with `SKILL.md` (spawn_agents_on_csv native): `analyze-with-file`, `brainstorm`, `brainstorm-with-file`, `clean`, `csv-wave-pipeline`, `debug-with-file`, `issue-discover`, `parallel-dev-cycle`, `project-documentation-workflow`, `review-cycle`, `roadmap-with-file`, `spec-generator`, `workflow-execute`, `workflow-lite-planex`, `workflow-plan`, `workflow-tdd-plan`, `workflow-test-fix-cycle`, `team-planex`, `team-coordinate`, `team-lifecycle-v4`, `team-issue`, `team-review`, `team-testing`, `team-quality-assurance`, `team-tech-debt`, `team-perf-opt`, `team-arch-opt`, `team-brainstorm`, `team-ultra-analyze`, `team-uidesign`, `team-ui-polish`, `team-ux-improve`, `team-visual-a11y`, `team-frontend`, `team-frontend-debug`, `team-interactive-craft`, `team-motion-design`, `team-roadmap-dev` Skills with `orchestrator.md` (phase-based): `investigate`, `security-audit`, `ship`, `memory-capture` ### Chain Definitions (task_type → skill sequence) > All `$skill-name` references below correspond to actual `.codex/skills/{skill-name}/` directories. > **[B]** = barrier skill (solo wave, coordinator analyzes artifacts after) | task_type | Chain name | Steps (skills, in order) | |-----------|-----------|--------------------------| | `bugfix-hotfix` | `bugfix.hotfix` | $workflow-lite-planex `--hotfix` |