maestro-help
Maestro Flow 命令帮助系统。搜索命令、浏览技能、工作流推荐、新手引导。Triggers on
git clone --depth 1 https://github.com/catlog22/maestro-flow /tmp/maestro-help && cp -r /tmp/maestro-help/.codex/skills/maestro-help ~/.claude/skills/maestro-helpSKILL.md
# Maestro Help
Maestro Flow 命令帮助系统,提供命令搜索、技能浏览、工作流推荐、新手引导功能。
## Trigger Conditions
- 关键词: "maestro-help", "帮助", "命令", "怎么用", "maestro 怎么用", "工作流", "skill", "workflow", "有哪些命令", "用什么命令"
- 场景: 询问命令用法、搜索命令、请求下一步建议、选择工作流、浏览 Skill/Agent 目录
- 斜杠: `/maestro-help`, `/maestro-help search <keyword>`, `/maestro-help skills`, `/maestro-help guide`
## Architecture Overview
```
┌──────────────────────────────────────────────────────────────────┐
│ Maestro Help (SKILL.md) — Orchestrator │
│ → Parse intent → Route to mode → Execute phase → Present │
└────────────────────────┬─────────────────────────────────────────┘
│
┌────────────────────┼────────────────────────┐
↓ ↓ ↓
┌──────────┐ ┌──────────────┐ ┌──────────┐
│ Phase 1 │ │ Phase 2 │ │ Phase 3 │
│ Parse │─────→│ Search & │────────→│ Workflow │
│ Intent │ │ Present │ │ Guide │
└──────────┘ └──────────────┘ └──────────┘
↑ ↗ │
└──────┘ ↓
(refine search) present guide
```
## Key Design Principles
1. **Catalog 驱动**: 所有查询基于 `index/catalog.json`,不做硬编码
2. **Guide 深度链接**: 命令详情链接到 `guide/` 目录中的参考文档
3. **上下文感知**: 根据项目状态(.workflow/ 是否存在、当前 Phase)调整推荐
4. **中英双语**: 命令名英文,说明和示例中文
## Data Source
Single source of truth: **[index/catalog.json](index/catalog.json)**
| Field | Purpose |
|-------|---------|
| `commands[]` | 64 个 slash 命令,含分类和描述 |
| `skills[]` | 23 个 Skill(含 10 个选装 scholar-*,标 `optional: true`),含分类和描述 |
| `agents[]` | 24 个 Agent,含分类和描述 |
| `cli_commands[]` | 21 个终端命令 |
| `guide_files[]` | 17 个 Guide 文档索引(planned,尚未创建) |
| `essential_commands[]` | 10 个核心命令(新手用) |
| `workflows` | 主干管线、Companion 轻量入口、Issue 闭环、初始化路径 |
## Operation Modes
### Mode 1: Command Search
**Triggers**: "搜索命令", "find command", "search", 命令名关键词
**Process**:
1. Read `Ref: phases/01-parse-intent.md` — 解析搜索意图
2. Query `catalog.json` commands[] + cli_commands[]
3. Filter by name, description, category
4. Present top 5 相关结果,含命令名、描述、分类
### Mode 2: Command Documentation
**Triggers**: "怎么用", "how to use", "详情", 具体命令名
**Process**:
1. Locate command in `catalog.json`
2. Read source file via `source` path(从 catalog 相对路径)
3. 若有对应 guide 文档,读取并提取相关段落
4. 提供上下文相关的用法示例
### Mode 3: Smart Recommendations
**Triggers**: "下一步", "what's next", "推荐", "继续"
**Process**:
1. 检测当前项目状态(.workflow/state.json)
2. 根据 workflows 配置推荐后续命令
3. Explain WHY 每个推荐适合当前状态
### Mode 4: Workflow Guide
**Triggers**: "工作流", "workflow", "怎么开始", "用什么流程"
**Process**:
1. Read `Ref: phases/03-workflow-guide.md`
2. 分析用户任务类型和复杂度
3. 推荐匹配的工作流(主干管线/Companion 轻量入口/Issue 闭环)
4. 给出具体命令序列
### Mode 5: Beginner Onboarding
**Triggers**: "新手", "getting started", "常用命令", "入门"
**Process**:
1. Query `catalog.json` essential_commands[]
2. 逐个展示核心命令的简要说明
3. 引导用户完成首次项目初始化
### Mode 6: Skill & Agent Browsing
**Triggers**: "skill", "agent", "技能", "有哪些 skill", "团队"
**Process**:
1. Read `Ref: phases/02-search-present.md`
2. Query `catalog.json` skills[] 或 agents[]
3. Filter by category
4. 呈现分类列表,含描述
### Mode 7: CLI Command Reference
**Triggers**: "终端命令", "CLI", "maestro 命令", "terminal"
**Process**:
1. Query `catalog.json` cli_commands[]
2. 按分类分组呈现
3. 含别名和常用选项
## Execution Flow
```
Input: $ARGUMENTS (free text)
Phase 1: Parse Intent
└─ Ref: phases/01-parse-intent.md
├─ 分析关键词确定 operation mode
├─ 提取搜索词 / 命令名 / 分类过滤
└─ Output: { mode, query, category?, context? }
Phase 2: Search & Present (Mode 1/2/3/6/7)
└─ Ref: phases/02-search-present.md
├─ 查询 catalog.json
├─ 按模式过滤和排序
├─ 读取 source 文件(Mode 2)
└─ Output: 格式化结果
Phase 3: Workflow Guide (Mode 4/5)
└─ Ref: phases/03-workflow-guide.md
├─ 检测项目状态
├─ 匹配工作流模板
├─ 生成推荐命令序列
└─ Output: 引导信息
```
**Phase Reference Documents** (read on-demand):
| Phase | Document | Purpose |
|-------|----------|---------|
| 1 | [phases/01-parse-intent.md](phases/01-parse-intent.md) | 意图解析和模式路由 |
| 2 | [phases/02-search-present.md](phases/02-search-present.md) | 搜索和呈现 |
| 3 | [phases/03-workflow-guide.md](phases/03-workflow-guide.md) | 工作流推荐和引导 |
## Input Processing
```
$ARGUMENTS → Parse:
├─ "search <keyword>" → Mode 1: Command Search
├─ 命令名 (如 "analyze") → Mode 2: Documentation
├─ "下一步" / "next" → Mode 3: Smart Recommendations
├─ "工作流" / "workflow" → Mode 4: Workflow Guide
├─ "新手" / "入门" → Mode 5: Beginner Onboarding
├─ "skill" / "agent" → Mode 6: Skill & Agent Browsing
├─ "CLI" / "终端" → Mode 7: CLI Reference
├─ 空参数 → Mode 5: Beginner Onboarding
└─ 其他自由文本 → Mode 1: Command Search (fuzzy)
```
## Command Catalog Quick Reference
### 上游起源 + 核心 (core)
> 裸名称为 first-tier step:经 `/maestro "<意图>"` 自动路由,或按 v3 receipt chain 直接执行:`session open` → fenced `session chain insert --command <step> --arg "<domain text>"` → fenced `run next`。birth packet 内嵌 prepare guidance,并返回 resolved `task` 与 structured `continuation`;`--input` 仅接受 sealed same-Session Artifact ID。`/` 前缀为独立命令。
| 命令 | 用途 |
|------|------|
| `/maestro` | 智能协调器,自动路由 |
| `/maestro-init` | 项目初始化 |
| `brainstorm` | 头脑风暴 — 发散探索,多角色创意 |
| `blueprint` | 正式规格文档化 — 7-phase 收敛规格链 |
| `roadmap` | 路线图编排 — 消费上游 context,纯 Milestone > Phase 分解 |
| `/maestro-companion` | 轻量任务直接执行 |
| `/maestro-overlay` | Overlay 管理 — 自然语言创建,或 `--amend` 从信号自动生成修正补丁 |
| `grill` | 压力测试 — 对计划或需求进行代码库现实性压力测试 |
| `/maestro-next` | 智能导航 — 检测状态并推荐下一步最优命令 |
| `/maestro-ralph --engine swarm` | Swarm 并行加速器 — 多 agent 并发执行 |
| `/maestro-ralph --engine universal` | 动态对抗工作流生成器 |
### 理解层 + 执行管线 (pipeline)
| 命令 | 用途 |
|------|------|
| `analyze` | 双层分析 — 宏观(文本参数)探索影响面 / 微观(数字参数)Phase 级深入 |
| `plan` | 任务规划 — 支持 `--from analyze:ANL-xxx` 直达 |
| `execute` | 任务执行 |
### 质量管线 (quality)
| 命令 | 用途 |
|------|------|
| `review` | 代码审查 |
| `auto-test` | 自动测试 |
| `test` | 业务测试 |
| `debRead-only code exploration via Bash + CLI semantic dual-source analysis, with schema-validated structured output.
Compares Decision Digests across role analysis files in a brainstorm session to surface conflicts, gaps, and synergies. Read-only — returns structured text for the orchestrator to apply.
Autonomous executor for non-interactive impeccable commands. Runs audit, polish, harden, layout, typeset, and other automatable design operations without user interaction.
Generates multi-file role analysis for a brainstorm session — analysis.md index + per-feature files + optional findings under {output_dir}/{role}/.
Resident pipeline supervisor agent. Message-driven lifecycle for cross-checkpoint quality observation and health monitoring.
Unified worker agent for team pipelines. Executes role-specific logic loaded from a role_spec file within a built-in task lifecycle (discover, execute, report).
UI design token management and prototype generation — W3C Design Tokens Format, state-based components, WCAG AA validation, responsive layout templates.
Evaluates technical topics, proposals, or decisions across multiple dimensions with evidence-based scoring and recommendations.