@dommaker/harness 0.9.0 → 0.10.0

package/dist/tools/definitions/definitions/std/governance/audit-task.yml

@@ -0,0 +1,178 @@
+# audit-task.yml - 任务审计步骤
+# ============================================================================
+# 类型: 治理类步骤
+# 执行方式: Agent（auditor 角色）
+# 用途: 审计特定任务的执行质量
+# 核心理念: 独立审计，发现异常，追溯原因
+# ============================================================================
+
+name: audit-task
+stage: govern
+description: 审计特定任务的执行质量
+category: governance
+version: 1.0.0
+
+# Agent 配置
+agent: codex
+temperature: 0.3  # 低温度，保证一致性
+
+# 工具引用
+tools:
+  - file-read
+  - git/log
+  - git/diff
+  - analysis/analyze-code
+
+# 输入定义
+inputs:
+  task_id:
+    type: string
+    description: 要审计的任务 ID
+    required: true
+  
+  project_path:
+    type: path
+    description: 项目路径
+    required: true
+  
+  audit_depth:
+    type: string
+    description: 审计深度
+    enum: [quick, standard, deep]
+    default: standard
+  
+  focus_areas:
+    type: array
+    description: 重点审计领域
+    default: [quality, process, compliance]
+    items:
+      type: string
+      enum: [quality, process, compliance, security, performance]
+
+# 审计检查项
+audit_checks:
+  quality:
+    - name: 代码质量
+      checks:
+        - 是否遵循编码规范
+        - 是否有充分的测试覆盖
+        - 是否有明显的 bug 或问题
+    - name: 文档完整性
+      checks:
+        - 是否更新了相关文档
+        - 注释是否清晰
+        - 提交信息是否规范
+  
+  process:
+    - name: 流程合规
+      checks:
+        - 是否按工作流步骤执行
+        - 是否跳过了必做步骤
+        - 是否通过了必要的审核
+    - name: 签署文档
+      checks:
+        - 是否有必要的签署记录
+        - 签署是否完整
+        - 是否有遗漏的审核环节
+  
+  compliance:
+    - name: 约束遵守
+      checks:
+        - 是否遵守约束级别要求
+        - 是否在权限范围内操作
+        - 是否有越权行为
+
+# Agent Prompt
+prompt: |
+  你是审计官（Auditor），独立于项目团队，负责监督。
+  
+  ## 你的职责
+  - 发现执行中的问题
+  - 追溯问题根源
+  - 评估严重程度
+  - 提出改进建议
+  
+  ## 你的立场
+  - 独立客观，不受项目进度影响
+  - 严谨细致，不放过细节
+  - 公正公平，不偏袒任何角色
+  
+  ## 禁止行为
+  - 参与项目开发
+  - 修改代码或文档
+  - 影响项目进度决策
+  
+  ## 审计任务
+  审计任务: {{task_id}}
+  项目路径: {{project_path}}
+  审计深度: {{audit_depth}}
+  重点领域: {{focus_areas}}
+  
+  ## 输出格式
+  
+  ```yaml
+  audit_result:
+    task_id: {{task_id}}
+    
+    findings:
+      - category: quality/process/compliance
+        severity: critical/major/minor/suggestion
+        description: 问题描述
+        evidence:
+          - 证据1
+          - 证据2
+        recommendation: 改进建议
+    
+    statistics:
+      issues_found: N
+      critical: N
+      major: N
+      minor: N
+      suggestion: N
+    
+    overall_assessment: pass/warn/fail
+    
+    impeachment_required: true/false
+    impeachment_reason: 如果需要弹劾，说明原因
+  ```
+
+# 输出定义
+outputs:
+  - name: audit_result
+    type: object
+    description: 审计结果
+  
+  - name: findings
+    type: array
+    description: 审计发现列表
+    items:
+      type: object
+      properties:
+        category:
+          type: string
+        severity:
+          type: string
+        description:
+          type: string
+        evidence:
+          type: array
+        recommendation:
+          type: string
+  
+  - name: overall_assessment
+    type: string
+    enum: [pass, warn, fail]
+  
+  - name: impeachment_required
+    type: boolean
+  
+  - name: impeachment_reason
+    type: string
+
+# 元数据
+metadata:
+  author: Agent Studio
+  tags:
+    - governance
+    - audit
+    - quality

package/dist/tools/definitions/definitions/std/governance/cast-vote.yml

@@ -0,0 +1,129 @@
+# cast-vote.yml - 投票步骤
+# ============================================================================
+# 类型: 治理类步骤
+# 执行方式: 内置处理器
+# 用途: 角色对进化建议/审计报告进行投票
+# 核心理念: 每人一票，Owner 有否决权
+# ============================================================================
+
+name: cast-vote
+stage: govern
+description: 角色对进化建议/审计报告进行投票
+category: governance
+version: 1.0.0
+
+# 执行配置
+execute:
+  type: builtin
+  handler: governance/cast-vote
+
+# 输入定义
+inputs:
+  voting_session_id:
+    type: string
+    description: 投票会话 ID
+    required: true
+  
+  voter_id:
+    type: string
+    description: 投票者 ID（角色 ID 或 'owner'）
+    required: true
+  
+  voter_type:
+    type: string
+    description: 投票者类型
+    enum: [role, user]
+    default: role
+  
+  vote:
+    type: string
+    description: 投票选择
+    enum: [approve, reject, abstain]
+    required: true
+  
+  reason:
+    type: string
+    description: 投票理由
+    required: false
+  
+  weight:
+    type: string
+    description: 投票权重
+    enum: [normal, heavy, veto]
+    default: normal
+
+# 投票规则
+voting_rules:
+  # 权重映射
+  weight_values:
+    normal: 1
+    heavy: 2
+    veto: 999999  # 否决权视为无限权重
+  
+  # 默认权重分配
+  default_weights:
+    auditor: normal
+    architect: normal
+    decider: normal
+    owner: veto
+  
+  # 禁止行为
+  forbidden:
+    - 不能投自己的票（角色不能对自己相关的建议投票）
+    - 不能重复投票
+    - 不能修改已投的票
+
+# 输出定义
+outputs:
+  - name: vote_recorded
+    type: boolean
+    description: 投票是否成功记录
+  
+  - name: vote_id
+    type: string
+    description: 投票记录 ID
+  
+  - name: session_status
+    type: string
+    description: 投票会话当前状态
+  
+  - name: current_statistics
+    type: object
+    description: 当前投票统计
+    properties:
+      approve_count:
+        type: number
+      reject_count:
+        type: number
+      abstain_count:
+        type: number
+      total_voters:
+        type: number
+      remaining_voters:
+        type: number
+
+# 错误处理
+errors:
+  VOTER_NOT_PARTICIPANT:
+    message: "投票者不在参与者列表中"
+    severity: error
+  
+  ALREADY_VOTED:
+    message: "该投票者已投过票"
+    severity: error
+  
+  SESSION_CLOSED:
+    message: "投票会话已关闭"
+    severity: error
+  
+  CONFLICT_OF_INTEREST:
+    message: "投票者与投票事项存在利益冲突"
+    severity: error
+
+# 元数据
+metadata:
+  author: Agent Studio
+  tags:
+    - governance
+    - voting
+    - decision-making

package/dist/tools/definitions/definitions/std/governance/collect-issues.yml

@@ -0,0 +1,16 @@
+# collect-issues.yml - 收集问题
+id: governance/collect-issues
+stage: govern
+name: 收集问题
+description: 从审核意见中收集问题列表
+
+inputs:
+  - name: opinions
+    type: array
+    required: true
+
+outputs:
+  - name: issues
+    type: array
+  - name: suggestions
+    type: array

package/dist/tools/definitions/definitions/std/governance/create-signed-document.yml

@@ -0,0 +1,23 @@
+# create-signed-document.yml - 创建签署文档
+id: governance/create-signed-document
+stage: govern
+name: 创建签署文档
+description: 创建带签名的审核文档
+
+inputs:
+  - name: document_type
+    type: string
+    required: true
+  - name: content
+    type: string
+    required: true
+  - name: signers
+    type: array
+    required: true
+  - name: metadata
+    type: object
+    required: false
+
+outputs:
+  - name: document
+    type: object

package/dist/tools/definitions/definitions/std/governance/create-voting-session.yml

@@ -0,0 +1,101 @@
+# create-voting-session.yml - 创建投票会话
+# ============================================================================
+# 类型: 治理类步骤
+# 执行方式: 内置处理器
+# 用途: 创建新的投票会话
+# ============================================================================
+
+name: create-voting-session
+stage: govern
+description: 创建新的投票会话
+category: governance
+version: 1.0.0
+
+# 执行配置
+execute:
+  type: builtin
+  handler: governance/create-voting-session
+
+# 输入定义
+inputs:
+  topic:
+    type: string
+    description: 投票主题
+    required: true
+  
+  description:
+    type: string
+    description: 详细描述
+    required: false
+  
+  related_item_id:
+    type: string
+    description: 关联的项目 ID（进化建议/审计报告）
+    required: true
+  
+  related_item_type:
+    type: string
+    description: 关联项目类型
+    enum: [evolution, audit, impeachment]
+    required: true
+  
+  participants:
+    type: array
+    description: 参与者列表
+    items:
+      type: object
+      properties:
+        id:
+          type: string
+        type:
+          type: string
+          enum: [role, user]
+        weight:
+          type: string
+          enum: [normal, heavy, veto]
+        required:
+          type: boolean
+    required: true
+  
+  config:
+    type: object
+    description: 投票配置
+    properties:
+      method:
+        type: string
+        enum: [simple_majority, absolute_majority, unanimous, super_majority]
+      threshold:
+        type: number
+      duration_minutes:
+        type: number
+      min_participants:
+        type: number
+      quorum:
+        type: number
+    required: true
+
+# 输出定义
+outputs:
+  - name: voting_session_id
+    type: string
+    description: 投票会话 ID
+  
+  - name: status
+    type: string
+    enum: [pending, voting, completed, cancelled]
+    description: 会话状态
+  
+  - name: voting_ends_at
+    type: number
+    description: 投票截止时间（时间戳）
+  
+  - name: participant_count
+    type: number
+    description: 参与者数量
+
+# 元数据
+metadata:
+  author: Agent Studio
+  tags:
+    - governance
+    - voting

package/dist/tools/definitions/definitions/std/governance/define-stance.sh

@@ -0,0 +1,243 @@
+#!/bin/bash
+# define-stance.sh - 立场注入工具
+# ============================================================================
+# 功能：根据立场 ID 返回对应的 prompt 和约束
+# 用途：三省六部制核心工具，引导 Agent 以特定立场思考
+# ============================================================================
+
+set -e
+
+# 参数
+STANCE_ID="${1:-}"
+TASK="${2:-}"
+INPUT_CONTENT="${3:-}"
+CONTEXT="${4:-}"
+
+# 颜色输出
+RED='\033[0;31m'
+GREEN='\033[0;32m'
+YELLOW='\033[1;33m'
+NC='\033[0m'
+
+log_info() { echo -e "${GREEN}[INFO]${NC} $1"; }
+log_error() { echo -e "${RED}[ERROR]${NC} $1"; }
+
+# 显示帮助
+show_help() {
+  cat << EOF
+立场注入工具
+
+用法:
+  $0 <stance_id> [task] [input_content] [context]
+
+参数:
+  stance_id     立场 ID（见下方列表）
+  task          当前任务描述
+  input_content 输入内容
+  context       上下文信息
+
+立场列表:
+
+决策类:
+  critic      挑刺者（评审专家）- 找问题、提风险
+  supporter   支持者（方案策划）- 找亮点、推方案
+  decider     决策者（项目负责人）- 平衡风险与收益
+
+执行类:
+  planner     规划者 - 制定计划、分解任务
+  executor    执行者（开发工程师）- 实现功能
+  tester      测试者 - 验证功能、发现 Bug
+
+专业类:
+  architect   架构师 - 设计系统架构
+  security    安全专家 - 安全评估
+  performance 性能专家 - 性能优化
+
+审计:
+  auditor     审计官 - 合规检查
+
+输出:
+  JSON 格式的立场信息和注入 prompt
+
+示例:
+  $0 critic "审核方案" "架构设计方案"
+  $0 architect "设计架构" "需求文档"
+EOF
+  exit 0
+}
+
+# 检查参数
+if [[ "$1" == "-h" || "$1" == "--help" || -z "$STANCE_ID" ]]; then
+  show_help
+fi
+
+# 立场定义
+declare -A STANCES
+
+# 决策类
+STANCES["critic"]='{
+  "id": "critic",
+  "name": "Critic",
+  "nameZh": "挑刺者",
+  "category": "decision",
+  "description": "评审专家，职责是找出问题和风险",
+  "injected_prompt": "你是评审专家（Critic），你的核心职责是挑刺和找问题。你需要在方案中发现潜在风险、设计缺陷、实现问题。你的立场是：宁可错杀一千，不可放过一个隐患。",
+  "forbidden_actions": ["批准方案", "忽视风险", "简化问题"],
+  "focus_areas": ["潜在问题", "风险评估", "边界条件", "失败场景"],
+  "review_criteria": ["完整性", "可行性", "安全性", "可维护性"]
+}'
+
+STANCES["supporter"]='{
+  "id": "supporter",
+  "name": "Supporter",
+  "nameZh": "支持者",
+  "category": "decision",
+  "description": "方案策划，职责是发现亮点和推动方案",
+  "injected_prompt": "你是方案策划（Supporter），你的核心职责是发现方案的亮点和价值。你需要找出方案的优势、创新点、可行性证据。你的立场是：寻找支持方案的理由，帮助方案通过审核。",
+  "forbidden_actions": ["否定方案", "忽视亮点", "制造障碍"],
+  "focus_areas": ["方案优势", "创新点", "价值证明", "成功案例"],
+  "review_criteria": ["价值", "创新性", "可行性", "性价比"]
+}'
+
+STANCES["decider"]='{
+  "id": "decider",
+  "name": "Decider",
+  "nameZh": "决策者",
+  "category": "decision",
+  "description": "项目负责人，职责是平衡风险与收益做出决策",
+  "injected_prompt": "你是决策者（Decider），你的核心职责是平衡风险与收益，做出最终决策。你需要综合考虑各方意见，评估资源约束、时间约束、团队能力。你的立场是：在充分信息基础上做出最优决策。",
+  "forbidden_actions": ["盲目批准", "情绪化决策", "忽视风险"],
+  "focus_areas": ["风险收益平衡", "资源约束", "时间约束", "团队能力"],
+  "review_criteria": ["风险可控", "资源充足", "时间合理", "能力匹配"]
+}'
+
+# 执行类
+STANCES["planner"]='{
+  "id": "planner",
+  "name": "Planner",
+  "nameZh": "规划者",
+  "category": "execution",
+  "description": "制定计划、分解任务",
+  "injected_prompt": "你是规划者（Planner），你的核心职责是制定执行计划和分解任务。你需要将大目标拆解为可执行的步骤，考虑依赖关系和并行可能。你的立场是：计划要清晰、完整、可执行。",
+  "forbidden_actions": ["忽略依赖", "模糊任务", "无时间估算"],
+  "focus_areas": ["任务分解", "依赖关系", "并行机会", "里程碑"],
+  "review_criteria": ["任务清晰", "依赖明确", "可执行性", "时间合理"]
+}'
+
+STANCES["executor"]='{
+  "id": "executor",
+  "name": "Executor",
+  "nameZh": "执行者",
+  "category": "execution",
+  "description": "开发工程师，职责是实现功能",
+  "injected_prompt": "你是执行者（Executor），你的核心职责是实现功能代码。你需要考虑代码质量、可维护性、性能。你的立场是：写出高质量、可维护的代码。",
+  "forbidden_actions": ["写烂代码", "忽略测试", "跳过文档"],
+  "focus_areas": ["代码质量", "可维护性", "性能", "测试覆盖"],
+  "review_criteria": ["代码规范", "测试通过", "文档完整", "性能达标"]
+}'
+
+STANCES["tester"]='{
+  "id": "tester",
+  "name": "Tester",
+  "nameZh": "测试者",
+  "category": "execution",
+  "description": "验证功能、发现 Bug",
+  "injected_prompt": "你是测试者（Tester），你的核心职责是验证功能和发现 Bug。你需要设计测试用例、执行测试、报告问题。你的立场是：确保功能正确、质量达标。",
+  "forbidden_actions": ["跳过测试", "忽视边界", "模糊报告"],
+  "focus_areas": ["功能验证", "边界测试", "性能测试", "回归测试"],
+  "review_criteria": ["功能正确", "覆盖完整", "Bug 清晰", "回归通过"]
+}'
+
+# 专业类
+STANCES["architect"]='{
+  "id": "architect",
+  "name": "Architect",
+  "nameZh": "架构师",
+  "category": "professional",
+  "description": "设计系统架构",
+  "injected_prompt": "你是架构师（Architect），你的核心职责是设计系统架构。你需要考虑可扩展性、可维护性、性能、安全。你的立场是：架构要稳定、灵活、易维护。",
+  "forbidden_actions": ["过度设计", "忽视约束", "技术炫技"],
+  "focus_areas": ["架构设计", "技术选型", "扩展性", "可维护性"],
+  "review_criteria": ["架构清晰", "技术合理", "扩展性好", "文档完整"]
+}'
+
+STANCES["security"]='{
+  "id": "security",
+  "name": "Security",
+  "nameZh": "安全专家",
+  "category": "professional",
+  "description": "安全评估",
+  "injected_prompt": "你是安全专家（Security），你的核心职责是评估系统安全性。你需要发现安全漏洞、评估风险、提出安全建议。你的立场是：安全第一，风险可控。",
+  "forbidden_actions": ["忽视漏洞", "弱化风险", "跳过评估"],
+  "focus_areas": ["安全漏洞", "风险评估", "合规检查", "安全建议"],
+  "review_criteria": ["无高危漏洞", "风险可控", "合规达标", "防护完善"]
+}'
+
+STANCES["performance"]='{
+  "id": "performance",
+  "name": "Performance",
+  "nameZh": "性能专家",
+  "category": "professional",
+  "description": "性能优化",
+  "injected_prompt": "你是性能专家（Performance），你的核心职责是优化系统性能。你需要发现性能瓶颈、提出优化方案、验证优化效果。你的立场是：性能要达标、资源要高效。",
+  "forbidden_actions": ["忽视瓶颈", "过度优化", "无基准测试"],
+  "focus_areas": ["性能瓶颈", "优化方案", "基准测试", "资源利用"],
+  "review_criteria": ["性能达标", "瓶颈消除", "资源高效", "稳定运行"]
+}'
+
+# 审计
+STANCES["auditor"]='{
+  "id": "auditor",
+  "name": "Auditor",
+  "nameZh": "审计官",
+  "category": "audit",
+  "description": "合规检查",
+  "injected_prompt": "你是审计官（Auditor），你的核心职责是检查合规性。你需要验证流程合规、文档完整、标准达标。你的立场是：合规是底线，必须严格执行。",
+  "forbidden_actions": ["放宽标准", "忽视流程", "跳过检查"],
+  "focus_areas": ["流程合规", "文档完整", "标准达标", "问题追踪"],
+  "review_criteria": ["流程合规", "文档完整", "标准达标", "问题闭环"]
+}'
+
+# 获取立场信息
+STANCE_JSON="${STANCES[$STANCE_ID]}"
+
+if [[ -z "$STANCE_JSON" ]]; then
+  log_error "未知的立场 ID: $STANCE_ID"
+  cat << EOF
+{
+  "success": false,
+  "error": "Unknown stance ID: $STANCE_ID",
+  "supported_stances": ["critic", "supporter", "decider", "planner", "executor", "tester", "architect", "security", "performance", "auditor"]
+}
+EOF
+  exit 1
+fi
+
+# 构建完整 prompt
+INJECTED_PROMPT=$(echo "$STANCE_JSON" | jq -r '.injected_prompt')
+
+FULL_PROMPT="$INJECTED_PROMPT"
+
+if [[ -n "$TASK" ]]; then
+  FULL_PROMPT="$FULL_PROMPT\n\n## 当前任务\n$TASK"
+fi
+
+if [[ -n "$INPUT_CONTENT" ]]; then
+  FULL_PROMPT="$FULL_PROMPT\n\n## 输入内容\n$INPUT_CONTENT"
+fi
+
+if [[ -n "$CONTEXT" ]]; then
+  FULL_PROMPT="$FULL_PROMPT\n\n## 上下文\n$CONTEXT"
+fi
+
+# 输出结果
+cat << EOF
+{
+  "success": true,
+  "stance_id": "$STANCE_ID",
+  "stance": $STANCE_JSON,
+  "injected_prompt": $(echo "$FULL_PROMPT" | jq -Rs .),
+  "forbidden_actions": $(echo "$STANCE_JSON" | jq '.forbidden_actions'),
+  "focus_areas": $(echo "$STANCE_JSON" | jq '.focus_areas')
+}
+EOF