feat: implement enterprise workflow with multi-backend support

## Overview
Complete implementation of enterprise-level workflow features including
multi-backend execution (Codex/Claude/Gemini), GitHub issue-to-PR automation,
hooks system, and comprehensive documentation.

## Major Changes

### 1. Multi-Backend Support (codeagent-wrapper)
- Renamed codex-wrapper → codeagent-wrapper
- Backend interface with Codex/Claude/Gemini implementations
- Multi-format JSON stream parser (auto-detects backend)
- CLI flag: --backend codex|claude|gemini (default: codex)
- Test coverage: 89.2%

**Files:**
- codeagent-wrapper/backend.go - Backend interface
- codeagent-wrapper/parser.go - Multi-format parser
- codeagent-wrapper/config.go - CLI parsing with backend selection
- codeagent-wrapper/executor.go - Process execution
- codeagent-wrapper/logger.go - Async logging
- codeagent-wrapper/utils.go - Utilities

### 2. GitHub Workflow Commands
- /gh-create-issue - Create structured issues via guided dialogue
- /gh-implement - Issue-to-PR automation with full dev lifecycle

**Files:**
- github-workflow/commands/gh-create-issue.md
- github-workflow/commands/gh-implement.md
- skills/codeagent/SKILL.md

### 3. Hooks System
- UserPromptSubmit hook for skill activation
- Pre-commit example with code quality checks
- merge_json operation in install.py for settings.json merging

**Files:**
- hooks/skill-activation-prompt.sh|.js
- hooks/pre-commit.sh
- hooks/hooks-config.json
- hooks/test-skill-activation.sh

### 4. Skills System
- skill-rules.json for auto-activation
- codeagent skill for multi-backend wrapper

**Files:**
- skills/skill-rules.json
- skills/codeagent/SKILL.md
- skills/codex/SKILL.md (updated)

### 5. Installation System
- install.py: Added merge_json operation
- config.json: Added "gh" module
- config.schema.json: Added op_merge_json schema

### 6. CI/CD
- GitHub Actions workflow for testing and building

**Files:**
- .github/workflows/ci.yml

### 7. Comprehensive Documentation
- Architecture overview with ASCII diagrams
- Codeagent-wrapper complete usage guide
- GitHub workflow detailed examples
- Hooks customization guide

**Files:**
- docs/architecture.md (21KB)
- docs/CODEAGENT-WRAPPER.md (9KB)
- docs/GITHUB-WORKFLOW.md (9KB)
- docs/HOOKS.md (4KB)
- docs/enterprise-workflow-ideas.md
- README.md (updated with doc links)

## Test Results
- All tests passing 
- Coverage: 89.2%
- Security scan: 0 issues (gosec)

## Breaking Changes
- codex-wrapper renamed to codeagent-wrapper
- Default backend: codex (documented in README)

## Migration Guide
Users with codex-wrapper installed should:
1. Run: python3 install.py --module dev --force
2. Update shell aliases if any

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
This commit is contained in:
swe-agent[bot]
2025-12-09 15:53:31 +08:00
parent 1533e08425
commit 3ef288bfaa
40 changed files with 5249 additions and 1406 deletions

407
docs/CODEAGENT-WRAPPER.md Normal file
View File

@@ -0,0 +1,407 @@
# Codeagent-Wrapper User Guide
Multi-backend AI code execution wrapper supporting Codex, Claude, and Gemini.
## Overview
`codeagent-wrapper` is a Go-based CLI tool that provides a unified interface to multiple AI coding backends. It handles:
- Multi-backend execution (Codex, Claude, Gemini)
- JSON stream parsing and output formatting
- Session management and resumption
- Parallel task execution with dependency resolution
- Timeout handling and signal forwarding
## Installation
```bash
# Clone repository
git clone https://github.com/cexll/myclaude.git
cd myclaude
# Install via install.py (includes binary compilation)
python3 install.py --module dev
# Or manual installation
cd codeagent-wrapper
go build -o ~/.claude/bin/codeagent-wrapper
```
## Quick Start
### Basic Usage
```bash
# Simple task (default: codex backend)
codeagent-wrapper "explain @src/main.go"
# With backend selection
codeagent-wrapper --backend claude "refactor @utils.ts"
# With HEREDOC (recommended for complex tasks)
codeagent-wrapper --backend gemini - <<'EOF'
Implement user authentication:
- JWT tokens
- Password hashing with bcrypt
- Session management
EOF
```
### Backend Selection
| Backend | Command | Best For |
|---------|---------|----------|
| **Codex** | `--backend codex` | General code tasks (default) |
| **Claude** | `--backend claude` | Complex reasoning, architecture |
| **Gemini** | `--backend gemini` | Fast iteration, prototyping |
## Core Features
### 1. Multi-Backend Support
```bash
# Codex (default)
codeagent-wrapper "add logging to @app.js"
# Claude for architecture decisions
codeagent-wrapper --backend claude - <<'EOF'
Design a microservices architecture for e-commerce:
- Service boundaries
- Communication patterns
- Data consistency strategy
EOF
# Gemini for quick prototypes
codeagent-wrapper --backend gemini "create React component for user profile"
```
### 2. File References with @ Syntax
```bash
# Single file
codeagent-wrapper "optimize @src/utils.ts"
# Multiple files
codeagent-wrapper "refactor @src/auth.ts and @src/middleware.ts"
# Entire directory
codeagent-wrapper "analyze @src for security issues"
```
### 3. Session Management
```bash
# First task
codeagent-wrapper "add validation to user form"
# Output includes: SESSION_ID: 019a7247-ac9d-71f3-89e2-a823dbd8fd14
# Resume session
codeagent-wrapper resume 019a7247-ac9d-71f3-89e2-a823dbd8fd14 - <<'EOF'
Now add error messages for each validation rule
EOF
```
### 4. Parallel Execution
Execute multiple tasks concurrently with dependency management:
```bash
codeagent-wrapper --parallel <<'EOF'
---TASK---
id: backend_1701234567
workdir: /project/backend
---CONTENT---
implement /api/users endpoints with CRUD operations
---TASK---
id: frontend_1701234568
workdir: /project/frontend
---CONTENT---
build Users page consuming /api/users
---TASK---
id: tests_1701234569
workdir: /project/tests
dependencies: backend_1701234567, frontend_1701234568
---CONTENT---
add integration tests for user management flow
EOF
```
**Parallel Task Format:**
- `---TASK---` - Starts task block
- `id: <unique_id>` - Required, use `<feature>_<timestamp>` format
- `workdir: <path>` - Optional, defaults to current directory
- `dependencies: <id1>, <id2>` - Optional, comma-separated task IDs
- `---CONTENT---` - Separates metadata from task content
**Features:**
- Automatic topological sorting
- Unlimited concurrency for independent tasks
- Error isolation (failures don't stop other tasks)
- Dependency blocking (skip if parent fails)
### 5. Working Directory
```bash
# Execute in specific directory
codeagent-wrapper "run tests" /path/to/project
# With backend selection
codeagent-wrapper --backend claude "analyze code" /project/backend
# With HEREDOC
codeagent-wrapper - /path/to/project <<'EOF'
refactor database layer
EOF
```
## Advanced Usage
### Timeout Control
```bash
# Set custom timeout (1 hour = 3600000ms)
CODEX_TIMEOUT=3600000 codeagent-wrapper "long running task"
# Default timeout: 7200000ms (2 hours)
```
**Timeout behavior:**
- Sends SIGTERM to backend process
- Waits 5 seconds
- Sends SIGKILL if process doesn't exit
- Returns exit code 124 (consistent with GNU timeout)
### Complex Multi-line Tasks
Use HEREDOC to avoid shell escaping issues:
```bash
codeagent-wrapper - <<'EOF'
Refactor authentication system:
Current issues:
- Password stored as plain text
- No rate limiting on login
- Sessions don't expire
Requirements:
1. Hash passwords with bcrypt
2. Add rate limiting (5 attempts/15min)
3. Session expiry after 24h
4. Add refresh token mechanism
Files to modify:
- @src/auth/login.ts
- @src/middleware/rateLimit.ts
- @config/session.ts
EOF
```
### Backend-Specific Features
**Codex:**
```bash
# Best for code editing and refactoring
codeagent-wrapper --backend codex - <<'EOF'
extract duplicate code in @src into reusable helpers
EOF
```
**Claude:**
```bash
# Best for complex reasoning
codeagent-wrapper --backend claude - <<'EOF'
review @src/payment/processor.ts for:
- Race conditions
- Edge cases
- Security vulnerabilities
EOF
```
**Gemini:**
```bash
# Best for fast iteration
codeagent-wrapper --backend gemini "add TypeScript types to @api.js"
```
## Output Format
Standard output includes parsed agent messages and session ID:
```
Agent response text here...
Implementation details...
---
SESSION_ID: 019a7247-ac9d-71f3-89e2-a823dbd8fd14
```
Error output (stderr):
```
ERROR: Error message details
```
Parallel execution output:
```
=== Parallel Execution Summary ===
Total: 3 | Success: 2 | Failed: 1
--- Task: backend_1701234567 ---
Status: SUCCESS
Session: 019a7247-ac9d-71f3-89e2-a823dbd8fd14
Implementation complete...
--- Task: frontend_1701234568 ---
Status: SUCCESS
Session: 019a7248-ac9d-71f3-89e2-a823dbd8fd14
UI components created...
--- Task: tests_1701234569 ---
Status: FAILED (exit code 1)
Error: dependency backend_1701234567 failed
```
## Exit Codes
| Code | Meaning |
|------|---------|
| 0 | Success |
| 1 | General error (missing args, no output) |
| 124 | Timeout |
| 127 | Backend command not found |
| 130 | Interrupted (Ctrl+C) |
| * | Passthrough from backend process |
## Environment Variables
| Variable | Default | Description |
|----------|---------|-------------|
| `CODEX_TIMEOUT` | 7200000 | Timeout in milliseconds |
## Troubleshooting
**Backend not found:**
```bash
# Ensure backend CLI is installed
which codex
which claude
which gemini
# Check PATH
echo $PATH
```
**Timeout too short:**
```bash
# Increase timeout to 4 hours
CODEX_TIMEOUT=14400000 codeagent-wrapper "complex task"
```
**Session ID not found:**
```bash
# List recent sessions (backend-specific)
codex history
# Ensure session ID is copied correctly
codeagent-wrapper resume <session_id> "continue task"
```
**Parallel tasks not running:**
```bash
# Check task format
# Ensure ---TASK--- and ---CONTENT--- delimiters are correct
# Verify task IDs are unique
# Check dependencies reference existing task IDs
```
## Integration with Claude Code
Use via the `codeagent` skill:
```bash
# In Claude Code conversation
User: Use codeagent to implement authentication
# Claude will execute:
codeagent-wrapper --backend codex - <<'EOF'
implement JWT authentication in @src/auth
EOF
```
## Performance Tips
1. **Use parallel execution** for independent tasks
2. **Choose the right backend** for the task type
3. **Keep working directory specific** to reduce context
4. **Resume sessions** for multi-step workflows
5. **Use @ syntax** to minimize file content in prompts
## Best Practices
1. **HEREDOC for complex tasks** - Avoid shell escaping nightmares
2. **Descriptive task IDs** - Use `<feature>_<timestamp>` format
3. **Absolute paths** - Avoid relative path confusion
4. **Session resumption** - Continue conversations with context
5. **Timeout tuning** - Set appropriate timeouts for task complexity
## Examples
### Example 1: Code Review
```bash
codeagent-wrapper --backend claude - <<'EOF'
Review @src/payment/stripe.ts for:
1. Security issues (API key handling, input validation)
2. Error handling (network failures, API errors)
3. Edge cases (duplicate charges, partial refunds)
4. Code quality (naming, structure, comments)
EOF
```
### Example 2: Refactoring
```bash
codeagent-wrapper --backend codex - <<'EOF'
Refactor @src/utils:
- Extract duplicate code into helpers
- Add TypeScript types
- Improve function naming
- Add JSDoc comments
EOF
```
### Example 3: Full-Stack Feature
```bash
codeagent-wrapper --parallel <<'EOF'
---TASK---
id: api_1701234567
workdir: /project/backend
---CONTENT---
implement /api/notifications endpoints with WebSocket support
---TASK---
id: ui_1701234568
workdir: /project/frontend
dependencies: api_1701234567
---CONTENT---
build Notifications component with real-time updates
---TASK---
id: tests_1701234569
workdir: /project
dependencies: api_1701234567, ui_1701234568
---CONTENT---
add E2E tests for notification flow
EOF
```
## Further Reading
- [Codex CLI Documentation](https://codex.docs)
- [Claude CLI Documentation](https://claude.ai/docs)
- [Gemini CLI Documentation](https://ai.google.dev/docs)
- [Architecture Overview](./architecture.md)

468
docs/GITHUB-WORKFLOW.md Normal file
View File

@@ -0,0 +1,468 @@
# GitHub Workflow Commands
Issue-to-PR automation with `codeagent-wrapper` integration.
## Overview
Two commands streamline GitHub workflows:
- `/gh-create-issue` - Create structured issues through guided dialogue
- `/gh-implement` - Implement issues with automated development lifecycle
## Prerequisites
```bash
# Install GitHub CLI
brew install gh
# Authenticate
gh auth login
# Verify
gh auth status
```
## Commands
### /gh-create-issue
Create structured GitHub issues through multi-round dialogue.
**Usage:**
```bash
/gh-create-issue "add user authentication"
```
**Process:**
1. **Initial Analysis** - Claude analyzes your brief description
2. **Clarification Questions** - Interactive Q&A to gather:
- Technical requirements
- Acceptance criteria
- Dependencies
- Priority/labels
3. **Issue Creation** - Generates structured issue with:
- Clear title
- Detailed description
- Acceptance criteria checklist
- Relevant labels
- Assignee (optional)
**Example Session:**
```
User: /gh-create-issue "add dark mode"
Claude: I'll help create a structured issue for dark mode. Let me gather some details.
1. Scope: Just theme toggle, or also persist user preference?
2. Which components need dark mode support?
3. Design system: Using CSS variables or styled-components?
User:
1. Toggle + persist in localStorage
2. All components in src/components
3. CSS variables
Claude: Creating issue...
✅ Issue created: #123 "Implement dark mode with theme persistence"
URL: https://github.com/user/repo/issues/123
```
**Generated Issue Format:**
```markdown
# Implement dark mode with theme persistence
## Description
Add dark mode support across all components with user preference persistence.
## Requirements
- [ ] Add theme toggle component
- [ ] Implement CSS variables for light/dark themes
- [ ] Persist theme preference in localStorage
- [ ] Update all components in src/components to support dark mode
- [ ] Add theme toggle to app header
## Acceptance Criteria
- [ ] User can toggle between light and dark themes
- [ ] Theme preference persists across sessions
- [ ] All UI components render correctly in both themes
- [ ] No flash of unstyled content on page load
## Technical Notes
- Use CSS custom properties
- Store preference as `theme: 'light' | 'dark'` in localStorage
- Add `data-theme` attribute to root element
Labels: enhancement, ui
```
---
### /gh-implement
Implement GitHub issue with full development lifecycle.
**Usage:**
```bash
/gh-implement 123
```
**Phases:**
#### Phase 1: Issue Analysis
```bash
# Fetches issue details
gh issue view 123 --json title,body,labels,comments
# Parses:
- Requirements
- Acceptance criteria
- Technical constraints
- Related discussions
```
#### Phase 2: Clarification (if needed)
Claude asks questions about:
- Implementation approach
- Architecture decisions
- Testing strategy
- Edge cases
#### Phase 3: Development
**Option A: Simple scope** - Direct `codeagent-wrapper` call:
```bash
codeagent-wrapper --backend codex - <<'EOF'
Implement dark mode toggle based on issue #123:
- Add ThemeToggle component
- Implement CSS variables
- Add localStorage persistence
EOF
```
**Option B: Complex scope** - Use `/dev` workflow:
```bash
/dev "implement issue #123: dark mode with theme persistence"
```
**Coverage requirement:** ≥90% test coverage enforced
#### Phase 4: Progress Updates
```bash
# After each milestone
gh issue comment 123 --body "✅ Completed: ThemeToggle component"
gh issue comment 123 --body "✅ Completed: CSS variables setup"
gh issue comment 123 --body "✅ Completed: localStorage persistence"
```
#### Phase 5: PR Creation
```bash
gh pr create \
--title "[#123] Implement dark mode with theme persistence" \
--body "Closes #123
## Changes
- Added ThemeToggle component
- Implemented light/dark CSS variables
- Added localStorage persistence
- Updated all components for theme support
## Testing
- Unit tests: ThemeToggle, theme utilities
- Integration tests: theme persistence across page loads
- Coverage: 92%"
```
**Output:**
```
✅ PR created: #124
URL: https://github.com/user/repo/pull/124
```
---
## Examples
### Example 1: Bug Fix
```bash
# Create issue
/gh-create-issue "login form doesn't validate email"
# Implement
/gh-implement 125
```
**Process:**
1. Analysis: Parse bug report, identify validation logic
2. Clarification: Confirm expected validation rules
3. Development: Fix validation, add tests
4. Updates: Comment with fix details
5. PR: Link to issue, show test coverage
---
### Example 2: Feature Development
```bash
# Create issue
/gh-create-issue "add export to CSV feature"
# Implement
/gh-implement 126
```
**Process:**
1. Analysis: Understand data structure, export requirements
2. Clarification: Which data fields? File naming? Encoding?
3. Development:
- Backend: CSV generation endpoint
- Frontend: Export button + download handler
- Tests: Unit + integration
4. Updates: Milestone comments (backend done, frontend done, tests done)
5. PR: Full feature description with screenshots
---
### Example 3: Refactoring
```bash
# Create issue
/gh-create-issue "refactor authentication module"
# Implement
/gh-implement 127
```
**Process:**
1. Analysis: Review current auth code, identify issues
2. Clarification: Scope (just refactor vs add features)?
3. Development:
- Modularize auth logic
- Extract reusable utilities
- Add missing tests
- Update documentation
4. Updates: Component-by-component progress
5. PR: Before/after comparison, test coverage improvement
---
## Workflow Integration
### With /dev Workflow
```bash
# Create issue first
/gh-create-issue "implement real-time notifications"
# Then implement with /dev
/gh-implement 128
# Claude will:
# 1. Analyze issue #128
# 2. Trigger /dev workflow internally
# 3. Execute with 90% coverage requirement
# 4. Post progress updates
# 5. Create PR
```
### With Parallel Tasks
For complex features, `/gh-implement` may use parallel execution:
```bash
# Internally executes:
codeagent-wrapper --parallel <<'EOF'
---TASK---
id: backend_notifications
workdir: /project/backend
---CONTENT---
implement notifications API with WebSocket
---TASK---
id: frontend_notifications
workdir: /project/frontend
dependencies: backend_notifications
---CONTENT---
build Notifications UI component
---TASK---
id: tests_notifications
workdir: /project
dependencies: backend_notifications, frontend_notifications
---CONTENT---
add E2E tests for notification flow
EOF
```
---
## Configuration
### Issue Templates
Create `.github/ISSUE_TEMPLATE/feature.md`:
```markdown
---
name: Feature Request
about: Suggest a new feature
labels: enhancement
---
## Description
<!-- Clear description of the feature -->
## Requirements
<!-- Specific requirements -->
## Acceptance Criteria
<!-- Checklist of criteria -->
```
### PR Templates
Create `.github/PULL_REQUEST_TEMPLATE.md`:
```markdown
## Related Issue
Closes #
## Changes
<!-- List of changes -->
## Testing
<!-- Test coverage and manual testing -->
## Screenshots (if applicable)
<!-- Before/after screenshots -->
```
---
## Best Practices
1. **Clear issue descriptions** - More context = better implementation
2. **Incremental commits** - Easier to review and rollback
3. **Test-driven** - Write tests before/during implementation
4. **Milestone updates** - Keep issue comments up-to-date
5. **Detailed PRs** - Explain why, not just what
---
## Troubleshooting
**Issue not found:**
```bash
# Verify issue exists
gh issue view 123
# Check repository
gh repo view
```
**PR creation failed:**
```bash
# Ensure branch is pushed
git push -u origin feature-branch
# Check if PR already exists
gh pr list --head feature-branch
```
**Authentication error:**
```bash
# Re-authenticate
gh auth login
# Check token scopes
gh auth status
```
---
## Advanced Usage
### Custom Labels
```bash
# Add labels during issue creation
gh issue create \
--title "Feature: dark mode" \
--body "..." \
--label "enhancement,ui,priority:high"
```
### Multiple Assignees
```bash
# Assign to team members
gh issue create \
--title "..." \
--assignee @user1,@user2
```
### Milestone Assignment
```bash
# Add to milestone
gh issue create \
--title "..." \
--milestone "v2.0"
```
---
## Integration with CI/CD
### Auto-close on merge
```yaml
# .github/workflows/pr-merge.yml
name: Close Issues on PR Merge
on:
pull_request:
types: [closed]
jobs:
close-issues:
if: github.event.pull_request.merged == true
runs-on: ubuntu-latest
steps:
- name: Close linked issues
run: gh issue close ${{ github.event.pull_request.number }}
```
### Coverage Check
```yaml
# .github/workflows/coverage.yml
name: Coverage Check
on: [pull_request]
jobs:
coverage:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- name: Run tests with coverage
run: go test -coverprofile=coverage.out ./...
- name: Check coverage threshold
run: |
coverage=$(go tool cover -func=coverage.out | grep total | awk '{print $3}' | sed 's/%//')
if (( $(echo "$coverage < 90" | bc -l) )); then
echo "Coverage $coverage% is below 90% threshold"
exit 1
fi
```
---
## Further Reading
- [GitHub CLI Manual](https://cli.github.com/manual/)
- [Codeagent-Wrapper Guide](./CODEAGENT-WRAPPER.md)
- [Hooks Documentation](./HOOKS.md)
- [Development Workflow](../README.md)

197
docs/HOOKS.md Normal file
View File

@@ -0,0 +1,197 @@
# Claude Code Hooks Guide
Hooks are shell scripts or commands that execute in response to Claude Code events.
## Available Hook Types
### 1. UserPromptSubmit
Runs after user submits a prompt, before Claude processes it.
**Use cases:**
- Auto-activate skills based on keywords
- Add context injection
- Log user requests
### 2. PostToolUse
Runs after Claude uses a tool.
**Use cases:**
- Validate tool outputs
- Run additional checks (linting, formatting)
- Log tool usage
### 3. Stop
Runs when Claude Code session ends.
**Use cases:**
- Cleanup temporary files
- Generate session reports
- Commit changes automatically
## Configuration
Hooks are configured in `.claude/settings.json`:
```json
{
"hooks": {
"UserPromptSubmit": [
{
"hooks": [
{
"type": "command",
"command": "$CLAUDE_PROJECT_DIR/hooks/skill-activation-prompt.sh"
}
]
}
],
"PostToolUse": [
{
"hooks": [
{
"type": "command",
"command": "$CLAUDE_PROJECT_DIR/hooks/post-tool-check.sh"
}
]
}
]
}
}
```
## Creating Custom Hooks
### Example: Pre-Commit Hook
**File:** `hooks/pre-commit.sh`
```bash
#!/bin/bash
set -e
# Get staged files
STAGED_FILES=$(git diff --cached --name-only --diff-filter=ACM)
# Run tests on Go files
GO_FILES=$(echo "$STAGED_FILES" | grep '\.go$' || true)
if [ -n "$GO_FILES" ]; then
go test ./... -short || exit 1
fi
# Validate JSON files
JSON_FILES=$(echo "$STAGED_FILES" | grep '\.json$' || true)
if [ -n "$JSON_FILES" ]; then
for file in $JSON_FILES; do
jq empty "$file" || exit 1
done
fi
echo "✅ Pre-commit checks passed"
```
**Register in settings.json:**
```json
{
"hooks": {
"PostToolUse": [
{
"hooks": [
{
"type": "command",
"command": "$CLAUDE_PROJECT_DIR/hooks/pre-commit.sh"
}
]
}
]
}
}
```
### Example: Auto-Format Hook
**File:** `hooks/auto-format.sh`
```bash
#!/bin/bash
# Format Go files
find . -name "*.go" -exec gofmt -w {} \;
# Format JSON files
find . -name "*.json" -exec jq --indent 2 . {} \; -exec mv {} {}.tmp \; -exec mv {}.tmp {} \;
echo "✅ Files formatted"
```
## Environment Variables
Hooks have access to:
- `$CLAUDE_PROJECT_DIR` - Project root directory
- `$PWD` - Current working directory
- All shell environment variables
## Best Practices
1. **Keep hooks fast** - Slow hooks block Claude Code
2. **Handle errors gracefully** - Return non-zero on failure
3. **Use absolute paths** - Reference `$CLAUDE_PROJECT_DIR`
4. **Make scripts executable** - `chmod +x hooks/script.sh`
5. **Test independently** - Run hooks manually first
6. **Document behavior** - Add comments explaining logic
## Debugging Hooks
Enable verbose logging:
```bash
# Add to your hook
set -x # Print commands
set -e # Exit on error
```
Test manually:
```bash
cd /path/to/project
./hooks/your-hook.sh
echo $? # Check exit code
```
## Built-in Hooks
This repository includes:
| Hook | File | Purpose |
|------|------|---------|
| Skill Activation | `skill-activation-prompt.sh` | Auto-suggest skills |
| Pre-commit | `pre-commit.sh` | Code quality checks |
## Disabling Hooks
Remove hook configuration from `.claude/settings.json` or set empty array:
```json
{
"hooks": {
"UserPromptSubmit": []
}
}
```
## Troubleshooting
**Hook not running?**
- Check `.claude/settings.json` syntax
- Verify script is executable: `ls -l hooks/`
- Check script path is correct
**Hook failing silently?**
- Add `set -e` to script
- Check exit codes: `echo $?`
- Add logging: `echo "debug" >> /tmp/hook.log`
## Further Reading
- [Claude Code Hooks Documentation](https://docs.anthropic.com/claude-code/hooks)
- [Bash Scripting Guide](https://www.gnu.org/software/bash/manual/)

502
docs/architecture.md Normal file
View File

@@ -0,0 +1,502 @@
# System Architecture
## Overview
Multi-agent AI development system with Claude Code as orchestrator and pluggable execution backends.
## High-Level Architecture
```
┌─────────────────────────────────────────────────────────────┐
│ User │
└──────────────────┬──────────────────────────────────────────┘
│ /dev, /gh-implement, etc.
┌─────────────────────────────────────────────────────────────┐
│ Claude Code (Orchestrator) │
│ ┌─────────────────────────────────────────────────────────┐│
│ │ - Planning & context gathering ││
│ │ - Requirements clarification ││
│ │ - Task breakdown ││
│ │ - Verification & reporting ││
│ └─────────────────────────────────────────────────────────┘│
└──────────────────┬──────────────────────────────────────────┘
│ via codeagent-wrapper
┌─────────────────────────────────────────────────────────────┐
│ Codeagent-Wrapper (Execution Layer) │
│ ┌──────────────────────────────────────────────────────────┤
│ │ Backend Interface │
│ ├──────────────┬──────────────┬──────────────┐ │
│ │ Codex │ Claude │ Gemini │ │
│ │ Backend │ Backend │ Backend │ │
│ └──────────────┴──────────────┴──────────────┘ │
│ │
│ ┌──────────────────────────────────────────────────────────┤
│ │ Features: │
│ │ - Multi-backend execution │
│ │ - JSON stream parsing │
│ │ - Session management │
│ │ - Parallel task execution │
│ │ - Timeout handling │
│ └──────────────────────────────────────────────────────────┘
└──────────────────┬──────────────────────────────────────────┘
│ CLI invocations
┌─────────────────────────────────────────────────────────────┐
│ AI CLI Backends │
│ ┌──────────────┬──────────────┬──────────────┐ │
│ │ Codex CLI │ Claude CLI │ Gemini CLI │ │
│ │ │ │ │ │
│ │ Code editing │ Reasoning │ Fast proto │ │
│ └──────────────┴──────────────┴──────────────┘ │
└─────────────────────────────────────────────────────────────┘
```
## Component Architecture
### 1. Orchestrator Layer (Claude Code)
**Responsibilities:**
- User interaction and requirements gathering
- Context analysis and exploration
- Task planning and breakdown
- Workflow coordination
- Verification and reporting
**Key Workflows:**
```
/dev
├── Requirements clarification (AskUserQuestion)
├── Codex analysis (Task tool → Explore agent)
├── Dev plan generation (Task tool → dev-plan-generator)
├── Parallel execution (codeagent-wrapper --parallel)
├── Coverage validation (≥90%)
└── Completion summary
/gh-implement <issue>
├── Issue analysis (gh issue view)
├── Clarification (if needed)
├── Development (codeagent-wrapper or /dev)
├── Progress updates (gh issue comment)
└── PR creation (gh pr create)
```
### 2. Execution Layer (Codeagent-Wrapper)
**Architecture:**
```go
Main Entry Point
- Parse CLI arguments
- Detect mode (new/resume/parallel)
- Select backend
Backend Selection
func SelectBackend(name string) Backend
CodexBackend ClaudeBackend GeminiBackend
Executor
func RunCodexTask(cfg *Config) (string, error)
1. Build command args via Backend.BuildArgs()
2. Start process with timeout
3. Stream stdout/stderr
4. Parse JSON stream via ParseJSONStream()
5. Extract session ID
6. Handle signals (SIGINT, SIGTERM)
Parser
func ParseJSONStream(r io.Reader) (string, string)
Detects format:
- Codex: {"type":"thread.started","thread_id":...}
- Claude: {"type":"...","subtype":"result"}
- Gemini: {"type":"...","role":"assistant"}
Extracts:
- Agent messages
- Session IDs
```
**Backend Interface:**
```go
type Backend interface {
Name() string
Command() string
BuildArgs(cfg *Config, targetArg string) []string
}
// Codex: codex e --skip-git-repo-check -C <workdir> --json <task>
// Claude: claude -p --dangerously-skip-permissions --output-format stream-json --verbose <task>
// Gemini: gemini -o stream-json -y -p <task>
```
**Key Files:**
- `main.go` - Entry point and orchestration
- `config.go` - CLI argument parsing
- `backend.go` - Backend interface and implementations
- `executor.go` - Process execution and stream handling
- `parser.go` - JSON stream parsing (multi-format)
- `logger.go` - Async logging with ring buffer
- `utils.go` - Helper functions
### 3. Hooks System
**Architecture:**
```
┌─────────────────────────────────────────────────────────┐
│ Claude Code Events │
│ UserPromptSubmit │ PostToolUse │ Stop │
└──────────────────┬──────────────────────────────────────┘
│ reads
┌─────────────────────────────────────────────────────────┐
│ .claude/settings.json │
│ { │
│ "hooks": { │
│ "UserPromptSubmit": [ │
│ { │
│ "hooks": [ │
│ { │
│ "type": "command", │
│ "command": "$CLAUDE_PROJECT_DIR/hooks/..." │
│ } │
│ ] │
│ } │
│ ] │
│ } │
│ } │
└──────────────────┬──────────────────────────────────────┘
│ executes
┌─────────────────────────────────────────────────────────┐
│ Hook Scripts │
│ ┌────────────────────────────────────────────────────┐ │
│ │ skill-activation-prompt.sh │ │
│ │ - Reads skills/skill-rules.json │ │
│ │ - Matches user prompt against triggers │ │
│ │ - Injects skill suggestions │ │
│ └────────────────────────────────────────────────────┘ │
│ ┌────────────────────────────────────────────────────┐ │
│ │ pre-commit.sh │ │
│ │ - Validates staged files │ │
│ │ - Runs tests │ │
│ │ - Formats code │ │
│ └────────────────────────────────────────────────────┘ │
└─────────────────────────────────────────────────────────┘
```
### 4. Skills System
**Structure:**
```
skills/
├── codex/
│ └── SKILL.md # Codex CLI integration
├── codeagent/
│ └── SKILL.md # Multi-backend wrapper
├── gemini/
│ └── SKILL.md # Gemini CLI integration
└── skill-rules.json # Auto-activation rules
```
**skill-rules.json Format:**
```json
{
"rules": [
{
"trigger": {
"pattern": "implement|build|create feature",
"type": "regex"
},
"skill": "codeagent",
"priority": 1,
"suggestion": "Use codeagent skill for code implementation"
}
]
}
```
## Data Flow
### Example: /dev Workflow
```
1. User: /dev "add user authentication"
2. Claude Code:
│ ├─ Clarifies requirements (AskUserQuestion)
│ ├─ Analyzes codebase (Explore agent)
│ └─ Generates dev-plan.md
3. Claude Code invokes: codeagent-wrapper --parallel <<EOF
---TASK---
id: auth_backend_1701234567
workdir: /project/backend
---CONTENT---
implement JWT authentication in @src/auth
---TASK---
id: auth_frontend_1701234568
workdir: /project/frontend
dependencies: auth_backend_1701234567
---CONTENT---
implement login form consuming /api/auth
---TASK---
id: auth_tests_1701234569
workdir: /project
dependencies: auth_backend_1701234567, auth_frontend_1701234568
---CONTENT---
add integration tests for auth flow
EOF
4. Codeagent-Wrapper:
│ ├─ Parses parallel config
│ ├─ Topological sort (resolves dependencies)
│ ├─ Executes tasks concurrently:
│ │ ├─ Task 1: codex e --json "implement JWT..."
│ │ ├─ Task 2: waits for Task 1, then codex e --json "implement login..."
│ │ └─ Task 3: waits for Tasks 1&2, then codex e --json "add tests..."
│ └─ Aggregates results
5. Claude Code:
│ ├─ Validates coverage (≥90%)
│ ├─ Runs final tests
│ └─ Reports summary
6. User receives:
✅ Authentication implemented
📊 Coverage: 92%
📁 Files modified: 8
🧪 Tests: 24 passed
```
## Module System
Installation system uses modular architecture:
```
config.json
├── dev module (enabled)
│ ├── merge_dir: dev-workflow → ~/.claude
│ ├── copy_file: memorys/CLAUDE.md
│ ├── copy_file: skills/codex/SKILL.md
│ └── run_command: install codeagent-wrapper binary
├── gh module (enabled)
│ ├── merge_dir: github-workflow → ~/.claude
│ ├── copy_file: skills/codeagent/SKILL.md
│ ├── copy_dir: hooks → ~/.claude/hooks
│ └── merge_json: hooks-config.json → settings.json
└── essentials module (enabled)
└── merge_dir: development-essentials → ~/.claude
```
**Installation Flow:**
```bash
python3 install.py --module dev,gh
1. Load config.json
2. Validate against config.schema.json
3. Select modules: dev, gh
4. Execute operations:
├─ dev:
│ ├─ Merge dev-workflow/commands → ~/.claude/commands
│ ├─ Copy CLAUDE.md → ~/.claude/CLAUDE.md
│ ├─ Copy codex skill → ~/.claude/skills/codex/
│ └─ Run install.sh (compile codeagent-wrapper)
└─ gh:
├─ Merge github-workflow/commands → ~/.claude/commands
├─ Copy codeagent skill → ~/.claude/skills/codeagent/
├─ Copy hooks → ~/.claude/hooks
└─ Merge hooks-config.json → ~/.claude/settings.json
5. Write installed_modules.json
6. Log to install.log
```
## Parallel Execution Engine
**Algorithm:**
```
1. Parse task config (---TASK--- delimited format)
2. Build dependency graph
3. Topological sort (detect cycles)
4. Execute in layers:
Layer 0: Tasks with no dependencies
Layer 1: Tasks depending only on Layer 0
Layer 2: Tasks depending on Layers 0-1
...
5. Within each layer: unlimited concurrency
6. On failure: skip dependent tasks, continue others
7. Aggregate results
```
**Example:**
```
Tasks:
A (no deps)
B (no deps)
C (depends on A)
D (depends on A, B)
E (depends on D)
Execution:
Layer 0: A, B (parallel)
Layer 1: C (waits for A), D (waits for A, B)
Layer 2: E (waits for D)
Timeline:
t=0: Start A, B
t=1: A completes
t=2: B completes, start C, D
t=3: C completes
t=4: D completes, start E
t=5: E completes
```
## Security Considerations
1. **No credential storage** - Uses existing CLI auth
2. **Sandbox execution** - Tasks run in specified workdir
3. **Timeout enforcement** - Prevents runaway processes
4. **Signal handling** - Graceful shutdown on Ctrl+C
5. **Input validation** - Sanitizes task configs
## Performance Characteristics
| Operation | Complexity | Notes |
|-----------|------------|-------|
| Backend selection | O(1) | Map lookup |
| Task parsing | O(n) | Linear scan |
| Topological sort | O(V+E) | Kahn's algorithm |
| Parallel execution | O(depth) | Depth of dependency graph |
| JSON parsing | O(n) | Streaming parser |
## Extensibility
### Adding New Backend
1. Implement `Backend` interface:
```go
type NewBackend struct{}
func (NewBackend) Name() string { return "new" }
func (NewBackend) Command() string { return "new-cli" }
func (NewBackend) BuildArgs(cfg *Config, targetArg string) []string {
return []string{"--json", targetArg}
}
```
2. Register in `config.go`:
```go
backendRegistry = map[string]Backend{
"codex": CodexBackend{},
"claude": ClaudeBackend{},
"gemini": GeminiBackend{},
"new": NewBackend{},
}
```
3. Add JSON format detection in `parser.go`:
```go
if hasKey(obj, "new_specific_field") {
// Parse new backend format
}
```
### Adding New Hook
1. Create script in `hooks/`:
```bash
#!/bin/bash
# hooks/my-custom-hook.sh
echo "Hook executed"
```
2. Register in `.claude/settings.json`:
```json
{
"hooks": {
"UserPromptSubmit": [
{
"hooks": [
{
"type": "command",
"command": "$CLAUDE_PROJECT_DIR/hooks/my-custom-hook.sh"
}
]
}
]
}
}
```
### Adding New Skill
1. Create `skills/my-skill/SKILL.md`:
```markdown
---
name: my-skill
description: My custom skill
---
# My Custom Skill
Usage instructions...
```
2. Add activation rule in `skills/skill-rules.json`:
```json
{
"rules": [
{
"trigger": {"pattern": "my keyword", "type": "regex"},
"skill": "my-skill",
"priority": 1
}
]
}
```
## Further Reading
- [Codeagent-Wrapper Guide](./CODEAGENT-WRAPPER.md)
- [GitHub Workflow Guide](./GITHUB-WORKFLOW.md)
- [Hooks Documentation](./HOOKS.md)
- [README](../README.md)

View File

@@ -0,0 +1,445 @@
# 企业级 Claude Code 工作流方案
基于 Anthropic 官方最佳实践、GitHub Copilot 企业级功能、以及 showcase 项目的研究整理。
## 实施状态
- ✅ codeagent-wrapper multi-backend
- ✅ /gh-create-issue command
- ✅ /gh-implement command
- ✅ Hooks + Skills activation
## 核心工作流矩阵
| 工作流 | 触发方式 | 核心能力 | 企业应用场景 |
|--------|----------|----------|--------------|
| `/gh-create-issue` | Command | 多轮对话 → 结构化 Issue | 需求澄清、Bug 报告标准化 |
| `/gh-implement` | Command | Issue → 开发 → PR | 自动化开发闭环 |
| `/code-review` | Hook (PR) | AI 审查 + 人工确认 | 代码质量把控 |
| `/incident-debug` | Command | 日志分析 → 根因定位 | 生产问题排查 |
| `/migration` | Command | 批量代码迁移 | 技术债务清理 |
| `/security-audit` | Hook/Scheduled | 安全扫描 + 修复建议 | 安全合规 |
| `/onboarding` | Command | 代码库问答 | 新人培训 |
---
## 1. GitHub Issue 全生命周期工作流
### 1.1 `/gh-create-issue` - 需求创建
```
用户输入 → 多轮澄清 → 结构化 Issue → gh issue create
```
**流程设计:**
```markdown
---
description: Create structured GitHub issue through multi-round dialogue
argument-hint: Brief description of what you need (e.g., "user authentication feature")
---
You are a Requirements Analyst. Help create a well-structured GitHub issue.
## Phase 1: Initial Understanding
Ask 2-3 targeted questions to understand:
- What problem does this solve? (Why)
- Who benefits from this? (Who)
- What's the expected outcome? (What)
## Phase 2: Technical Scoping
Based on answers, clarify:
- Acceptance criteria (testable conditions)
- Technical constraints
- Dependencies on other features/teams
- Priority and urgency
## Phase 3: Issue Generation
Generate issue with structure:
- **Title**: [Type] Brief description
- **Problem Statement**: Why this matters
- **Proposed Solution**: High-level approach
- **Acceptance Criteria**: Checkbox list
- **Technical Notes**: Implementation hints
- **Labels**: auto-suggest based on content
## Phase 4: Confirmation & Creation
Show preview → User confirms → `gh issue create`
```
### 1.2 `/gh-implement` - Issue 实现
```
gh issue view → 理解 + 沟通 → /dev 开发 → gh issue comment → gh pr create
```
**流程设计:**
```markdown
---
description: Implement GitHub issue with full development lifecycle
argument-hint: Issue number (e.g., "123")
---
## Phase 1: Issue Analysis
1. `gh issue view $ARGUMENTS --json title,body,labels,comments`
2. Parse requirements and acceptance criteria
3. Identify affected files via codebase exploration
## Phase 2: Clarification (if needed)
If ambiguous, use AskUserQuestion to clarify:
- Implementation approach choices
- Scope boundaries
- Testing requirements
## Phase 3: Development
Invoke /dev workflow with parsed requirements:
- Codex analysis
- Task breakdown
- Parallel execution
- Coverage validation (≥90%)
## Phase 4: Progress Updates
After each milestone:
`gh issue comment $ARGUMENTS --body "✅ Completed: [milestone]"`
## Phase 5: PR Creation
`gh pr create --title "[#$ARGUMENTS] ..." --body "Closes #$ARGUMENTS\n\n..."`
```
---
## 2. 代码审查工作流
### 2.1 PR 自动审查 Hook
**触发点:** PR 创建或更新时
```json
{
"hooks": {
"PostToolUse": [
{
"matcher": "Bash(gh pr create:*)",
"hooks": [{
"type": "command",
"command": "$CLAUDE_PROJECT_DIR/.claude/hooks/auto-review.sh"
}]
}
]
}
}
```
**审查维度(参考 Anthropic 博客):**
- 代码风格一致性
- 潜在 bug 和边界条件
- 安全漏洞OWASP Top 10
- 性能影响
- 文档完整性
- 测试覆盖率
### 2.2 `/review-pr` Command
```markdown
---
description: Comprehensive PR review with actionable feedback
argument-hint: PR number or URL
---
1. Fetch PR details: `gh pr view $ARGUMENTS --json files,commits,body`
2. Read changed files with context (±50 lines)
3. Analyze against:
- Repository coding standards (CLAUDE.md)
- Security best practices
- Performance implications
- Test coverage
4. Generate review with:
- Summary of changes
- 🟢 Approved / 🟡 Changes Requested / 🔴 Blocked
- Specific line comments
- Suggested improvements
5. Post review: `gh pr review $ARGUMENTS --body "..." [--approve|--request-changes]`
```
---
## 3. 生产问题排查工作流
### 3.1 `/incident-debug`
```markdown
---
description: Debug production incidents from logs and traces
argument-hint: Error message, log file path, or incident ID
---
## Phase 1: Context Gathering
- Parse provided logs/error messages
- Search codebase for related code paths
- Check recent deployments: `gh release list --limit 5`
## Phase 2: Root Cause Analysis
Use Codex for deep analysis:
- Stack trace interpretation
- Data flow tracing
- Dependency chain analysis
## Phase 3: Solution Proposal
- Immediate mitigation steps
- Long-term fix plan
- Regression test suggestions
## Phase 4: Documentation
Generate incident report:
- Timeline
- Root cause
- Impact assessment
- Resolution steps
- Prevention measures
```
---
## 4. 大规模迁移工作流
### 4.1 `/migration` - 批量代码迁移
**适用场景:**
- 框架升级React 17→18, Vue 2→3
- API 版本迁移
- 依赖库替换
- 代码模式重构
```markdown
---
description: Batch code migration with validation
argument-hint: Migration type and scope (e.g., "React class to hooks in src/components")
---
## Phase 1: Scope Analysis
1. Use Codex to identify all affected files
2. Generate migration task list (file by file)
3. Estimate complexity per file
## Phase 2: Parallel Execution (Headless Mode)
For each file, run:
```bash
claude -p "Migrate $FILE from [old] to [new]. Verify with tests." \
--allowedTools Edit Bash(npm test:*)
```
## Phase 3: Validation
- Run full test suite
- Type checking
- Lint verification
## Phase 4: Report
- Success/failure per file
- Manual review required files
- Rollback instructions
```
---
## 5. 安全审计工作流
### 5.1 `/security-audit`
```markdown
---
description: Security vulnerability scanning and remediation
---
## Scan Categories
1. **Dependency vulnerabilities**: `npm audit` / `pip-audit`
2. **SAST**: Code pattern analysis for OWASP Top 10
3. **Secrets detection**: Hardcoded credentials
4. **Configuration**: Insecure defaults
## Output Format
- Severity: Critical/High/Medium/Low
- Location: File:Line
- Description: What's wrong
- Remediation: How to fix
- Auto-fix available: Yes/No
## Auto-remediation
For auto-fixable issues:
1. Generate fix via Codex
2. Run tests
3. Create PR with security label
```
---
## 6. 新人培训工作流
### 6.1 Codebase Q&AAnthropic 推荐)
直接使用 Claude Code 进行代码库问答,无需特殊配置:
**常见问题类型:**
- "这个项目的架构是什么?"
- "如何添加新的 API 端点?"
- "日志系统是怎么工作的?"
- "这个函数为什么这样设计?"(结合 git history
### 6.2 `/onboarding` Command
```markdown
---
description: Interactive codebase onboarding for new team members
---
## Phase 1: Overview
- Read README, CLAUDE.md, package.json
- Summarize tech stack and architecture
## Phase 2: Key Flows
For each major feature:
- Entry point
- Data flow
- Key files
## Phase 3: Development Setup
- Environment setup steps
- Common commands
- Testing workflow
## Phase 4: Q&A Mode
"Ask me anything about this codebase!"
```
---
## 7. codeagent-wrapper 多后端架构
### 设计方案
```go
// codeagent-wrapper architecture
type AgentBackend interface {
Name() string
Execute(ctx context.Context, task TaskSpec, timeout int) TaskResult
HealthCheck() error
}
type CodexBackend struct{} // OpenAI Codex
type ClaudeBackend struct{} // Claude CLI (claude -p)
type GeminiBackend struct{} // Gemini API
// 命令行接口
// codeagent-wrapper [--backend=codex|claude|gemini] "task" [workdir]
// codeagent-wrapper --parallel --backend=claude < tasks.txt
```
### 后端选择策略
| 任务类型 | 推荐后端 | 原因 |
|----------|----------|------|
| 代码生成/重构 | Codex | 代码专精 |
| 复杂推理/规划 | Claude | 推理能力强 |
| 快速原型 | Gemini | 速度快、成本低 |
| 并行批量任务 | 混合 | 负载均衡 |
---
## 8. Hooks + Skills 协作模式
### 推荐配置
```json
{
"hooks": {
"UserPromptSubmit": [
{
"hooks": [{
"type": "command",
"command": "$CLAUDE_PROJECT_DIR/.claude/hooks/skill-activation-prompt.sh"
}]
}
],
"PostToolUse": [
{
"matcher": "Edit|MultiEdit|Write",
"hooks": [{
"type": "command",
"command": "$CLAUDE_PROJECT_DIR/.claude/hooks/post-tool-tracker.sh"
}]
},
{
"matcher": "Bash(gh pr create:*)",
"hooks": [{
"type": "command",
"command": "$CLAUDE_PROJECT_DIR/.claude/hooks/auto-review-trigger.sh"
}]
}
],
"Stop": [
{
"hooks": [
{"type": "command", "command": "$CLAUDE_PROJECT_DIR/.claude/hooks/test-runner.sh"},
{"type": "command", "command": "$CLAUDE_PROJECT_DIR/.claude/hooks/coverage-check.sh"}
]
}
]
}
}
```
### skill-rules.json 扩展
```json
{
"skills": {
"gh-workflow": {
"type": "domain",
"enforcement": "suggest",
"priority": "high",
"promptTriggers": {
"keywords": ["issue", "pr", "pull request", "github", "gh"],
"intentPatterns": ["(create|implement|review).*?(issue|pr|pull)"]
}
},
"incident-response": {
"type": "domain",
"enforcement": "suggest",
"priority": "critical",
"promptTriggers": {
"keywords": ["error", "bug", "incident", "production", "debug", "crash"],
"intentPatterns": ["(fix|debug|investigate).*?(error|bug|issue)"]
}
}
}
}
```
---
## 9. 实施优先级建议
### Phase 1: 基础设施1-2 周)
1. ✅ codeagent-wrapper 已完成
2. 🔄 codeagent-wrapper 多后端改造
3. 🆕 基础 hooks 配置
### Phase 2: 核心工作流2-3 周)
1. `/gh-create-issue` command
2. `/gh-implement` command
3. `/code-review` command
### Phase 3: 高级功能3-4 周)
1. skill-rules.json + activation hook
2. `/migration` 批量迁移
3. `/security-audit` 安全审计
### Phase 4: 企业级增强
1. 多 Claude 实例协作
2. CI/CD 集成headless mode
3. 监控和分析仪表板
---
## 参考资料
- [Anthropic Claude Code Best Practices](https://www.anthropic.com/engineering/claude-code-best-practices)
- [GitHub Copilot Coding Agent](https://docs.github.com/en/copilot/using-github-copilot/using-copilot-coding-agent-to-work-on-tasks)
- [claude-code-infrastructure-showcase](https://github.com/hellogithub/claude-code-infrastructure-showcase)