Files
Claude-Code-Workflow/.claude/skills/team-command-designer/phases/03-command-generation.md
catlog22 6054a01b8f feat: add CLI fallback for MCP calls in team commands
- Implemented CLI fallback using `ccw team` for various team command operations in `execute.md`, `plan.md`, `review.md`, `spec-analyst.md`, `spec-coordinate.md`, `spec-discuss.md`, `spec-reviewer.md`, `spec-writer.md`, and `test.md`.
- Updated command generation templates to include CLI fallback examples.
- Enhanced validation checks to ensure CLI fallback sections are present.
- Added quality standards for CLI fallback in team command design.
- Introduced a new `GlobalGraphExpander` class for expanding search results with cross-directory relationships.
- Added tests for `GlobalGraphExpander` to verify functionality and score decay factors.
2026-02-13 12:05:48 +08:00

365 lines
10 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
# Phase 3: Command Generation
Generate the team command .md file using template and pattern analysis results.
## Objective
- Apply command template with role-specific content
- Generate complete YAML front matter
- Generate message bus section
- Generate 5-phase implementation with code
- Generate error handling table
- Output final command file
## Input
- Dependency: `role-config.json` (Phase 1), `applicable-patterns.json` (Phase 2)
- Template: `templates/command-template.md`
## Execution Steps
### Step 1: Load Inputs
```javascript
const config = JSON.parse(Read(`${workDir}/role-config.json`))
const patterns = JSON.parse(Read(`${workDir}/applicable-patterns.json`))
const template = Read(`${skillDir}/templates/command-template.md`)
// Read most similar command for code reference
// Note: reference may be in a folder (e.g. .claude/commands/team/folder/cmd.md) if config.output_folder is set
const refCommand = Read(`.claude/commands/team/${patterns.similar_to.primary}.md`)
```
### Step 2: Generate YAML Front Matter
```javascript
const frontMatter = `---
name: ${config.role_name}
description: ${config.description_cn}
argument-hint: ""
allowed-tools: ${config.allowed_tools.join(', ')}
group: team
---`
```
### Step 3: Generate Message Bus Section
```javascript
const messageBusSection = `## Message Bus
Every SendMessage **before**, must call \`mcp__ccw-tools__team_msg\` to log:
\`\`\`javascript
mcp__ccw-tools__team_msg({ operation: "log", team: teamName, from: "${config.role_name}", to: "coordinator", type: "<type>", summary: "<summary>" })
\`\`\`
### Supported Message Types
| Type | Direction | Trigger | Description |
|------|-----------|---------|-------------|
${config.message_types.map(mt =>
`| \`${mt.type}\` | ${config.role_name} -> coordinator | ${mt.trigger} | ${mt.description || mt.trigger} |`
).join('\n')}
### Examples
\`\`\`javascript
${config.message_types.filter(mt => mt.type !== 'error').map(mt =>
`// ${mt.trigger}
mcp__ccw-tools__team_msg({ operation: "log", team: teamName, from: "${config.role_name}", to: "coordinator", type: "${mt.type}", summary: "${mt.trigger}" })`
).join('\n\n')}
\`\`\`
### CLI 回退
\`mcp__ccw-tools__team_msg\` MCP 不可用时,使用 \`ccw team\` CLI 作为等效回退:
\\\`\\\`\\\`javascript
// 回退: 将 MCP 调用替换为 Bash CLI参数一一对应
Bash(\\\`ccw team log --team "\${teamName}" --from "${config.role_name}" --to "coordinator" --type "${config.message_types[0]?.type || 'result'}" --summary "<摘要>" --json\\\`)
\\\`\\\`\\\`
**参数映射**: \`team_msg(params)\`\`ccw team log --team <team> --from ${config.role_name} --to coordinator --type <type> --summary "<text>" [--ref <path>] [--data '<json>'] [--json]\``
```
### Step 4: Generate Phase Implementations
```javascript
// Phase 1: Task Discovery (standard for all roles)
const phase1 = `### Phase 1: Task Discovery
\`\`\`javascript
// Find assigned ${config.task_prefix}-* tasks
const tasks = TaskList()
const myTasks = tasks.filter(t =>
t.subject.startsWith('${config.task_prefix}-') &&
t.owner === '${config.role_name}' &&
t.status === 'pending' &&
t.blockedBy.length === 0
)
if (myTasks.length === 0) return // idle
const task = TaskGet({ taskId: myTasks[0].id })
TaskUpdate({ taskId: task.id, status: 'in_progress' })
\`\`\``
// Phase 2: Context Loading (adapted by responsibility type)
const phase2Templates = {
"Read-only analysis": `### Phase 2: Context Loading
\`\`\`javascript
// Load plan for criteria reference
const planPathMatch = task.description.match(/\\.workflow\\/\\.team-plan\\/[^\\s]+\\/plan\\.json/)
let plan = null
if (planPathMatch) {
try { plan = JSON.parse(Read(planPathMatch[0])) } catch {}
}
// Get changed files
const changedFiles = Bash(\`git diff --name-only HEAD~1 2>/dev/null || git diff --name-only --cached\`)
.split('\\n').filter(Boolean)
// Read file contents for analysis
const fileContents = {}
for (const file of changedFiles.slice(0, 20)) {
try { fileContents[file] = Read(file) } catch {}
}
\`\`\``,
"Code generation": `### Phase 2: Task & Plan Loading
\`\`\`javascript
// Extract plan path from task description
const planPathMatch = task.description.match(/\\.workflow\\/\\.team-plan\\/[^\\s]+\\/plan\\.json/)
if (!planPathMatch) {
SendMessage({ type: "message", recipient: "coordinator",
content: \`Cannot find plan.json path in task \${task.subject}\`,
summary: "Plan path not found" })
return
}
const plan = JSON.parse(Read(planPathMatch[0]))
// Load task files from .task/ directory
const planTasks = plan.task_ids.map(id =>
JSON.parse(Read(\`\${planPathMatch[0].replace('plan.json', '')}.task/\${id}.json\`))
)
\`\`\``,
"Orchestration": `### Phase 2: Context & Complexity Assessment
\`\`\`javascript
// Assess task complexity
function assessComplexity(desc) {
let score = 0
if (/refactor|architect|restructure|module|system/.test(desc)) score += 2
if (/multiple|across|cross/.test(desc)) score += 2
if (/integrate|api|database/.test(desc)) score += 1
if (/security|performance/.test(desc)) score += 1
return score >= 4 ? 'High' : score >= 2 ? 'Medium' : 'Low'
}
const complexity = assessComplexity(task.description)
// Load related context
const projectTech = Bash(\`test -f .workflow/project-tech.json && cat .workflow/project-tech.json || echo "{}"\`)
\`\`\``,
"Validation": `### Phase 2: Environment Detection
\`\`\`javascript
// Detect relevant tools/frameworks
// (customize based on specific validation domain)
const changedFiles = Bash(\`git diff --name-only HEAD~1 2>/dev/null || git diff --name-only --cached\`)
.split('\\n').filter(Boolean)
// Load context based on validation type
\`\`\``
}
const phase2 = phase2Templates[config.responsibility_type]
// Phase 3: Core Work (role-specific, provides skeleton)
const phase3 = `### Phase 3: ${patterns.phase_structure.phase3}
\`\`\`javascript
// Core ${config.role_name} logic
// TODO: Implement based on role requirements
// Reference: .claude/commands/team/${patterns.similar_to.primary}.md Phase 3
${config.adaptive_routing ? `
// Complexity-adaptive execution
if (complexity === 'Low') {
// Direct execution
} else {
// Delegate to sub-agent
Task({
subagent_type: "universal-executor",
run_in_background: false,
description: "${config.role_name} work",
prompt: \`
## Task
\${task.description}
## MANDATORY FIRST STEPS
1. Read: .workflow/project-tech.json (if exists)
2. Read: .workflow/project-guidelines.json (if exists)
## Expected Output
\${expectedFormat}
\`
})
}` : `// Direct execution for all tasks`}
\`\`\``
// Phase 4: Validation/Summary
const phase4 = `### Phase 4: ${patterns.phase_structure.phase4}
\`\`\`javascript
// Validate/summarize results
// TODO: Implement based on role requirements
// Reference: .claude/commands/team/${patterns.similar_to.primary}.md Phase 4
\`\`\``
// Phase 5: Report + Loop (standard for all roles)
const phase5 = `### Phase 5: Report to Coordinator
\`\`\`javascript
// Log message before SendMessage
mcp__ccw-tools__team_msg({
operation: "log", team: teamName,
from: "${config.role_name}", to: "coordinator",
type: "${config.message_types[0]?.type || config.role_name + '_complete'}",
summary: \`${config.task_prefix} complete: \${task.subject}\`
})
SendMessage({
type: "message",
recipient: "coordinator",
content: \`## ${config.display_name} Results
**Task**: \${task.subject}
**Status**: \${resultStatus}
### Summary
\${resultSummary}
### Details
\${resultDetails}\`,
summary: \`${config.task_prefix} complete\`
})
// Mark task completed
TaskUpdate({ taskId: task.id, status: 'completed' })
// Check for next task
const nextTasks = TaskList().filter(t =>
t.subject.startsWith('${config.task_prefix}-') &&
t.owner === '${config.role_name}' &&
t.status === 'pending' &&
t.blockedBy.length === 0
)
if (nextTasks.length > 0) {
// Continue with next task -> back to Phase 1
}
\`\`\``
```
### Step 5: Generate Error Handling Table
```javascript
const errorTable = `## Error Handling
| Scenario | Resolution |
|----------|------------|
| No ${config.task_prefix}-* tasks available | Idle, wait for coordinator assignment |
| Plan/Context file not found | Notify coordinator, request location |
${config.adaptive_routing ? '| Sub-agent failure | Retry once, then fallback to direct execution |\n' : ''}| Max iterations exceeded | Report to coordinator, suggest intervention |
| Critical issue beyond scope | SendMessage fix_required to coordinator |
| Unexpected error | Log error via team_msg, report to coordinator |`
```
### Step 6: Assemble Final Command File
```javascript
const commandContent = `${frontMatter}
# Team ${config.display_name} Command (/${config.skill_path})
## Overview
Team ${config.role_name} role command. Operates as a teammate within an Agent Team, responsible for ${config.responsibility_type.toLowerCase()}.
**Core capabilities:**
- Task discovery from shared team task list (${config.task_prefix}-* tasks)
${patterns.core_patterns.includes('pattern-5-complexity-adaptive') ? '- Complexity-adaptive routing (Low -> direct, Medium/High -> agent)\n' : ''}- ${config.responsibility_type}-specific processing
- Structured result reporting to coordinator
## Role Definition
**Name**: \`${config.role_name}\`
**Responsibility**: ${patterns.phase_structure.phase2} -> ${patterns.phase_structure.phase3} -> ${patterns.phase_structure.phase5}
**Communication**: SendMessage to coordinator only
${messageBusSection}
## Execution Process
\`\`\`
Phase 1: Task Discovery
├─ TaskList to find unblocked ${config.task_prefix}-* tasks
├─ TaskGet to read full task details
└─ TaskUpdate to mark in_progress
Phase 2: ${patterns.phase_structure.phase2}
Phase 3: ${patterns.phase_structure.phase3}
Phase 4: ${patterns.phase_structure.phase4}
Phase 5: ${patterns.phase_structure.phase5}
├─ team_msg log + SendMessage results
├─ TaskUpdate completed
└─ Check for next ${config.task_prefix}-* task
\`\`\`
## Implementation
${phase1}
${phase2}
${phase3}
${phase4}
${phase5}
${errorTable}
`
Write(`${workDir}/${config.role_name}.md`, commandContent)
```
## Output
- **File**: `{role-name}.md`
- **Format**: Markdown
- **Location**: `{workDir}/{role-name}.md`
## Quality Checklist
- [ ] YAML front matter includes all required fields
- [ ] `group: team` is present
- [ ] Message bus section has team_msg examples
- [ ] All 5 phases are present with implementation code
- [ ] Task lifecycle follows standard pattern
- [ ] Error handling table is present
- [ ] SendMessage always preceded by team_msg
- [ ] Role communicates only with coordinator
## Next Phase
-> [Phase 4: Integration Verification](04-integration-verification.md)