1
name: "基础助手"
2
id: "basic-assistant"
3
version: "1.0.0"
4
description: "一个通用的 AI 助手，可以回答各种问题"
5

6
# 角色定义
7
role:
8
  name: "通用助手"
9
  expertise: ["通用问答", "文本处理", "信息检索"]
10
  personality: "友好、专业、耐心"
11

12
# 系统提示
13
systemPrompt: |
14
  你是一个 AI 助手，可以帮助用户解决各种问题。
15
  - 回答要准确、简洁
16
  - 不确定时坦诚告知
17
  - 保持友好专业的态度
18

19
# 模型配置
20
model:
21
  provider: "openai"
22
  model: "gpt-4o-mini"
23
  temperature: 0.7
24
  maxTokens: 1000
25

26
# 技能配置
27
skills:
28
  enabled:
29
    - "baidu-search"
30
    - "weather"
31
    - "calculator"
32
  disabled: []
33

34
# 工具配置
35
tools:
36
  enabled:
37
    - "file-read"
38
    - "web-fetch"
39
  disabled: []
40

41
# 上下文配置
42
context:
43
  maxMessages: 10
44
  maxTokens: 4000
45
  retainStrategy: "recent"
46

47
# 限制和权限
48
restrictions:
49
  allowedUsers: []
50
  allowedChannels: []
51
  rateLimit:
52
    messagesPerMinute: 20
53
    burstLimit: 5

1
routes:
2
  - pattern: ".*"
3
    agent: "basic-assistant"
4
    priority: 1

1
name: "代码审查专家"
2
id: "code-reviewer"
3
version: "1.0.0"
4
description: "专业的代码审查 Agent，专注于代码质量、安全性和最佳实践"
5

6
role:
7
  name: "代码审查专家"
8
  expertise: [
9
    "代码质量",
10
    "安全性检查",
11
    "性能优化",
12
    "设计模式",
13
    "测试覆盖"
14
  ]
15
  personality: "严格、细致、建设性"
16

17
systemPrompt: |
18
  你是一个代码审查专家。审查代码时请关注：
19
  1. 代码质量和可读性
20
  2. 安全漏洞和潜在风险
21
  3. 性能优化建议
22
  4. 遵循最佳实践
23
  5. 提供具体的改进建议
24

25
  输出格式：
26
  - 🔴 严重问题
27
  - 🟡 改进建议
28
  - ✅ 做得好的地方
29

30
model:
31
  provider: "openai"
32
  model: "gpt-4o"
33
  temperature: 0.3
34
  maxTokens: 2000
35

36
skills:
37
  enabled:
38
    - "code-analysis"
39
    - "security-scan"
40
    - "test-generator"
41
    - "documentation"
42

43
tools:
44
  enabled:
45
    - "file-read"
46
    - "git-diff"
47
    - "eslint"
48
    - "sonarqube"
49

50
context:
51
  maxMessages: 20
52
  maxTokens: 6000
53
  retainStrategy: "hybrid"
54

55
restrictions:
56
  allowedUsers: ["dev-team"]
57
  rateLimit:
58
    messagesPerMinute: 10

1
name: "技术写作专家"
2
id: "tech-writer"
3
version: "1.0.0"
4
description: "专业的技术文档撰写 Agent，擅长将复杂技术内容转化为清晰易懂的文档"
5

6
role:
7
  name: "技术写作专家"
8
  expertise: [
9
    "技术文档撰写",
10
    "API 文档",
11
    "用户手册",
12
    "教程编写",
13
    "Markdown 格式"
14
  ]
15
  personality: "清晰、准确、有条理"
16

17
systemPrompt: |
18
  你是一个技术写作专家。撰写文档时请注意：
19
  1. 使用清晰简洁的语言
20
  2. 遵循最佳写作实践
21
  3. 包含适当的示例和代码
22
  4. 使用标准 Markdown 格式
23
  5. 面向目标读者调整内容深度
24

25
  输出格式：
26
  - 标题使用 ##
27
  - 重要概念加粗
28
  - 代码使用代码块
29
  - 列表使用项目符号
30
  - 分节清晰
31

32
model:
33
  provider: "openai"
34
  model: "gpt-4o-mini"
35
  temperature: 0.6
36
  maxTokens: 3000
37

38
skills:
39
  enabled:
40
    - "research"
41
    - "summarization"
42
    - "formatting"
43
    - "translation"
44

45
tools:
46
  enabled:
47
    - "web-fetch"
48
    - "file-read"
49
    - "spell-check"

1
name: "数据分析专家"
2
id: "data-analyst"
3
version: "1.0.0"
4
description: "专业的数据分析 Agent，擅长数据处理、可视化和洞察发现"
5

6
role:
7
  name: "数据分析专家"
8
  expertise: [
9
    "数据清洗",
10
    "统计分析",
11
    "数据可视化",
12
    "机器学习",
13
    "报告生成"
14
  ]
15
  personality: "严谨、数据驱动、洞察力强"
16

17
systemPrompt: |
18
  你是一个数据分析专家。处理数据时：
19
  1. 理解数据结构和含义
20
  2. 选择合适的分析方法
21
  3. 提供有价值的洞察
22
  4. 生成清晰的图表
23
  5. 解释结果含义
24

25
  工作流程：
26
  1. 数据加载和探索
27
  2. 数据清洗和预处理
28
  3. 统计分析和建模
29
  4. 可视化呈现
30
  5. 洞察和建议
31

32
model:
33
  provider: "openai"
34
  model: "gpt-4o"
35
  temperature: 0.2
36
  maxTokens: 2500
37

38
skills:
39
  enabled:
40
    - "data-processing"
41
    - "statistical-analysis"
42
    - "visualization"
43
    - "machine-learning"
44
    - "reporting"
45

46
tools:
47
  enabled:
48
    - "pandas"
49
    - "matplotlib"
50
    - "plotly"
51
    - "scikit-learn"
52
    - "excel-reader"

1
name: "研究团队"
2
id: "research-team"
3
version: "1.0.0"
4
description: "一个多 Agent 协作团队，用于复杂的研究任务"
5

6
team:
7
  leader: "research-coordinator"
8
  members:
9
    - agent: "research-coordinator"
10
      role: "协调员"
11
      responsibilities:
12
        - "任务分配"
13
        - "进度跟踪"
14
        - "结果整合"
15
    - agent: "data-collector"
16
      role: "数据收集员"
17
      responsibilities:
18
        - "数据搜索"
19
        - "信息筛选"
20
        - "资源整理"
21
    - agent: "analyst"
22
      role: "分析师"
23
      responsibilities:
24
        - "数据分析"
25
        - "模式识别"
26
        - "洞察生成"
27
    - agent: "writer"
28
      role: "技术写手"
29
      responsibilities:
30
        - "报告撰写"
31
        - "文档生成"
32
        - "总结提炼"
33

34
workflow:
35
  steps:
36
    - step: 1
37
      agent: "research-coordinator"
38
      action: "understand_request"
39
    - step: 2
40
      agent: "research-coordinator"
41
      action: "plan_research"
42
    - step: 3
43
      agent: "data-collector"
44
      action: "collect_data"
45
    - step: 4
46
      agent: "analyst"
47
      action: "analyze_data"
48
    - step: 5
49
      agent: "writer"
50
      action: "generate_report"
51
    - step: 6
52
      agent: "research-coordinator"
53
      action: "review_and_finalize"
54

55
communication:
56
  enabled: true
57
  channels:
58
    - "internal-message"
59
    - "shared-context"
60
  protocol: "request-response"

1
dependencies:
2
  research-coordinator:
3
    requires:
4
      - "data-collector"
5
      - "analyst"
6
      - "writer"
7

8
  data-collector:
9
    dependsOn: []
10
    outputsTo: ["analyst"]
11

12
  analyst:
13
    dependsOn: ["data-collector"]
14
    outputsTo: ["writer"]
15

16
  writer:
17
    dependsOn: ["analyst"]
18
    outputsTo: ["research-coordinator"]

1
routes:
2
  - pattern: ".* 代码.*"
3
    agent: "code-reviewer"
4
    priority: 2
5
    examples:
6
      - "审查这段代码"
7
      - "代码质量检查"
8

9
  - pattern: ".* 文档.*|.* 撰写.*|.* 写作.*"
10
    agent: "tech-writer"
11
    priority: 2
12
    examples:
13
      - "撰写 API 文档"
14
      - "写一个教程"
15

16
  - pattern: ".* 数据.*|.* 分析.*|.* 统计.*"
17
    agent: "data-analyst"
18
    priority: 2
19
    examples:
20
      - "分析这组数据"
21
      - "数据可视化"
22

23
  - pattern: ".*"
24
    agent: "basic-assistant"
25
    priority: 1  # 默认路由

1
intentRoutes:
2
  - intent: "code_review"
3
    agent: "code-reviewer"
4
    keywords: ["代码", "审查", "bug", "调试"]
5
    confidenceThreshold: 0.7
6

7
  - intent: "documentation"
8
    agent: "tech-writer"
9
    keywords: ["文档", "手册", "教程", "指南"]
10
    confidenceThreshold: 0.7
11

12
  - intent: "data_analysis"
13
    agent: "data-analyst"
14
    keywords: ["数据", "分析", "统计", "图表"]
15
    confidenceThreshold: 0.7

1
contextRoutes:
2
  - condition: "user.role == 'developer'"
3
    agent: "code-reviewer"
4
    priority: 3
5

6
  - condition: "conversation.topic == 'documentation'"
7
    agent: "tech-writer"
8
    priority: 3
9

10
  - condition: "conversation.hasData == true"
11
    agent: "data-analyst"
12
    priority: 3

1
cache:
2
  enabled: true
3
  strategies:
4
    - type: "response"
5
      ttl: 3600
6
      maxSize: 1000
7
    - type: "context"
8
      ttl: 1800
9
      maxSize: 500
10
    - type: "skill"
11
      ttl: 7200
12
      maxSize: 2000

1
batching:
2
  enabled: true
3
  maxBatchSize: 5
4
  batchTimeout: 1000
5
  mergeStrategy: "compact"
6
  priorityLevels:
7
    - level: "high"
8
      batchSize: 1
9
    - level: "normal"
10
      batchSize: 3
11
    - level: "low"
12
      batchSize: 5

1
resources:
2
  limits:
3
    memory: "512M"
4
    cpu: "0.5"
5
    maxConcurrentTasks: 10
6
  reservations:
7
    memory: "128M"
8
    cpu: "0.1"
9
    minConcurrentTasks: 2

1
monitoring:
2
  agentMetrics:
3
    enabled: true
4
    metrics:
5
      - "response_time"
6
      - "token_usage"
7
      - "error_rate"
8
      - "success_rate"
9
      - "user_satisfaction"
10
    alertThresholds:
11
      response_time: 5000
12
      error_rate: 0.05
13
      token_usage: 10000

1
logging:
2
  level: "info"
3
  format: "json"
4
  outputs:
5
    - type: "console"
6
      enabled: true
7
    - type: "file"
8
      enabled: true
9
      path: "/var/log/openclaw/agents.log"
10
      maxSize: "100M"
11
      maxFiles: 10
12
    - type: "elasticsearch"
13
      enabled: false
14
      url: "http://localhost:9200"

1
audit:
2
  enabled: true
3
  events:
4
    - "agent_created"
5
    - "agent_modified"
6
    - "agent_deleted"
7
    - "agent_invoked"
8
    - "skill_called"
9
    - "error_occurred"
10
  retention:
11
    days: 90
12
    compress: true

1
security:
2
  permissions:
3
    read:
4
      - "file-system"
5
      - "database"
6
    write:
7
      - "file-system"
8
      - "database"
9
    execute:
10
      - "skills"
11
      - "tools"
12

13
  restrictions:
14
    forbiddenPaths:
15
      - "/etc"
16
      - "/root"
17
      - "/var/log"
18
    maxFileSize: 10485760  # 10MB
19
    allowedDomains:
20
      - "*.github.com"
21
      - "*.stackoverflow.com"

1
filters:
2
  input:
3
    enabled: true
4
    rules:
5
      - type: "pii"
6
        action: "redact"
7
      - type: "malicious"
8
        action: "block"
9
      - type: "spam"
10
        action: "quarantine"
11

12
  output:
13
    enabled: true
14
    rules:
15
      - type: "sensitive"
16
        action: "encrypt"
17
      - type: "confidential"
18
        action: "restrict"

1
describe('Basic Agent', () => {
2
  test('should respond to greeting', async () => {
3
    const agent = loadAgent('basic-assistant');
4
    const response = await agent.process(' 你好 ');
5
    expect(response).toContain(' 你好 ');
6
  });
7

8
  test('should handle unknown queries', async () => {
9
    const agent = loadAgent('basic-assistant');
10
    const response = await agent.process('xyz');
11
    expect(response).toContain(' 不确定 ');
12
  });
13
});

1
describe('Research Team Integration', () => {
2
  test('should complete full research workflow', async () => {
3
    const team = loadTeam('research-team');
4
    const result = await team.execute(' 研究 AI 的发展趋势 ');
5
    expect(result).toHaveProperty('report');
6
    expect(result).toHaveProperty('insights');
7
  });
8
});