CodeReview/config.yaml at main · crcde8ug/CodeReview · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
# AI Code Review Agent Configuration Example
# Copy this file to config.yaml and customize as needed

llm:
  # 一键切换模型：修改 provider 字段即可
  # - "deepseek" → 使用 DeepSeek (默认)
  # - "zhipuai" → 使用 ZhipuAI (GLM-4)
  # - "openai" → 使用 OpenAI
  # - "dashscope" → 使用阿里云 DashScope (Qwen)
  provider: "deepseek"  # Options: "openai", "deepseek", "zhipuai"

  # 模型名称（根据 provider 自动选择默认值）
  # DeepSeek: "deepseek-chat"
  # ZhipuAI: "glm-4.6"
  # OpenAI: "gpt-4", "gpt-3.5-turbo" 等
  # OpenRouter: "deepseek/deepseek-chat-v3-0324:free", "anthropic/claude-3.5-sonnet" 等
  model: "deepseek-chat"

  # API Key (可选，可通过环境变量设置)
  # 环境变量优先级: LLM_API_KEY > provider-specific keys (DEEPSEEK_API_KEY, ZHIPUAI_API_KEY)
  api_key: null  # 填入你的 OpenRouter API key，或通过 LLM_API_KEY 环境变量设置

  # Base URL (仅用于 OpenAI 兼容的 API，如 DeepSeek / OpenRouter)
  # ZhipuAI 不需要此配置
  base_url: "https://api.deepseek.com"

  # Temperature 参数 (0.0 - 2.0)
  temperature: 0

# 快速切换示例：
# 切换到 ZhipuAI: 将 provider 改为 "zhipuai"，model 改为 "glm-4"
# 切换到 OpenAI: 将 provider 改为 "openai"，model 改为 "gpt-4"
#
# 环境变量覆盖（最高优先级）：
# export LLM_PROVIDER=zhipuai  # 覆盖配置文件中的 provider
# export LLM_MODEL=glm-4       # 覆盖配置文件中的 model
# export LLM_API_KEY=your-key  # 覆盖配置文件中的 api_key
# export DASHSCOPE_API_KEY=your-key  # DashScope 专用 key（provider=dashscope 时读取）

system:
  workspace_root: "."           # Root directory of the workspace
  assets_dir: "assets_cache"    # Directory for storing built assets
  timeout_seconds: 600           # Maximum time for analysis (10 minutes)
  max_concurrent_llm_requests: 2
  max_expert_rounds: 22         # Maximum rounds for expert analysis (circuit breaker)
  max_expert_tool_calls: 20

  # ===== Noise control: file path filtering =====
  # 默认启用内置排除（锁文件/生成物/二进制/媒体/日志等），可用 include/exclude 进行覆盖
  path_filter_enabled: true
  path_filter_include_globs: []     # 例如 ["**/*.svg"] 强制把 svg 纳入审查
  path_filter_exclude_globs: []     # 例如 ["**/vendor/**"] 额外排除

  # ===== Reporter threshold =====
  confidence_threshold: 0.6
  # 可选：按风险类型设置更严格阈值（例如降低 Robustness 误报）
  # confidence_threshold_by_risk_type:
  #   Robustness_Boundary_Conditions: 0.75

  # ===== Manager gating / budgeting =====
  manager_anchor_window: 5          # 仅保留落在变更行±N 的风险
  manager_drop_unanchored: true
  manager_unanchored_confidence: 0.2
  manager_max_work_items_total: 15
  manager_max_items_per_file: 3
  # 建议：对 Robustness 做小配额，优先留给 Concurrency/Authorization/Lifecycle
  manager_max_items_per_risk_type:
    Robustness_Boundary_Conditions: 4
    Concurrency_Timing_Correctness: 4
    Authorization_Data_Exposure: 3
    Lifecycle_State_Consistency: 3
    Intent_Semantic_Consistency: 2
    Syntax_Static_Errors: 10
  manager_risk_type_weights: {}         # 为空则使用内置默认（降低 Robustness 权重）
  manager_severity_weights: {}          # 为空则使用内置默认（error>warning>info）
  manager_merge_line_window: 5
  manager_merge_jaccard: 0.75

  # ===== Expert calibration =====
  expert_confidence_clamp_on_budget_stop: 0.55

  # ===== Harness: Verify Loop =====
  harness_verify_loop_enabled: true
  harness_verify_confidence_floor: 0.3

  # ===== Harness: Eval Gate =====
  harness_eval_gate_enabled: true

  # ===== Harness: Progressive Context =====
  harness_progressive_context_enabled: true