道层 Skill 索引
Z2 条目↑ Z1 版块 ↓ Z3 全文
57 条判断类方法引导 · 按子类型分区 · 无配图
道层 skill 是判断类方法引导：57 条按子类型分区，不配独立图（配图需求由 W1b 的 16 张 dao 策展图承担），正文逐字投影直接可读。
道-方法 (53)
| 名称 | 一句话 | 真源路径 |
|---|---|---|
| ai_agent_cli_guide | 判断 CLI Agent 相比原始 API 何时更优并给出选型对照 | rules/skills/ai_agent_cli_guide.md |
| bestpractice_ai_debugging_diagnosis | 判断 AI 改不好代码时先查人类侧根因的决策树 | rules/skills/bestpractice_ai_debugging_diagnosis.md |
| bestpractice_ai_product_design | 判断 AI 产品为何应突破线性聊天、支持非线性知识工作 | rules/skills/bestpractice_ai_product_design.md |
| bestpractice_ai_programming_mindset | 判断 AI 编程 70% 问题的根因是反馈环断裂与成功标准缺失 | rules/skills/bestpractice_ai_programming_mindset.md |
| bestpractice_interview_evaluation | 判断 AI 时代面试应考 Trait 而非 Skill，并识别作弊 | rules/skills/bestpractice_interview_evaluation.md |
| bestpractice_skill_writing_guide | 判断 skill 文件应重结果确定性、避免写成机械脚本 | rules/skills/bestpractice_skill_writing_guide.md |
| bestpractice_staged_approach | 判断复杂自动化任务应分采集、处理、确认三阶段 | rules/skills/bestpractice_staged_approach.md |
| bestpractice_temporal_info_verification | 判断何时需主动验证可能超 knowledge cutoff 的信息 | rules/skills/bestpractice_temporal_info_verification.md |
| workflow_knowledge_flywheel | 判断知识工程应以笨数据+笨方法+笨模型迭代提纯 | rules/skills/workflow_knowledge_flywheel.md |
| workflow_real_task_test_case_design | 判断真实条件测试应沿消费链设计 case 与 oracle | rules/skills/workflow_real_task_test_case_design.md |
| workflow_real_task_test_case_design | redirect stub：指向正式版 ../workflow_real_task_test_case_design.md | rules/skills/drafts/workflow_real_task_test_case_design.md |
| workflow_test_design_audit | 判断测试文档/流程缺哪些卡点与历史测试遗漏 | rules/skills/workflow_test_design_audit.md |
| workflow_version_evolution | 判断对外承诺变更该打 X/Y 还是不升级版本号 | rules/skills/workflow_version_evolution.md |
| bestpractice_agent_defect_diagnosis | 判断 E2E 失败时区分 Agent 行为与 Skill 文档缺陷 | rules/skills/drafts/bestpractice_agent_defect_diagnosis.md |
| bestpractice_architecture_doc_design | 判断复杂系统架构文档应多源对比建唯一真源 | rules/skills/drafts/bestpractice_architecture_doc_design.md |
| bestpractice_deprecation_and_archive | 判断退役归档 vs 旧格式迁移的治理入口 | rules/skills/bestpractice_deprecation_and_archive.md |
| workflow_agent_communication_protocol | 判断多 agent 信息隔离、交接与禁读协议 | rules/skills/drafts/workflow_agent_communication_protocol.md |
| workflow_buildout_real_session_eval | 判断真实 Session 下验证 buildout 工具效果的方法 | rules/skills/drafts/workflow_buildout_real_session_eval.md |
| workflow_complexity_drift_detection | 判断何时部署检测门及检测器误差信号是否有效 | rules/skills/drafts/workflow_complexity_drift_detection.md |
| workflow_continuous_task_advancement | 判断完成声明可信度并驱动缺口补齐而非停在自评 | rules/skills/drafts/workflow_continuous_task_advancement.md |
| workflow_deferred_verification | 判断运行中任务应先标待核验、结束后回填证据 | rules/skills/drafts/workflow_deferred_verification.md |
| workflow_design_elegance | 判断设计应让单一元素覆盖多相容需求、反过度工程化 | rules/skills/drafts/workflow_design_elegance.md |
| workflow_error_recording | 判断多 provider 运行时错误应按 taxonomy 记录并复用 | rules/skills/drafts/workflow_error_recording.md |
| workflow_iterative_expanding_research | 判断多轮调研应以覆盖矩阵空白格驱动扩张 | rules/skills/drafts/workflow_iterative_expanding_research.md |
| workflow_landing_to_production | 判断 draft 是否真 landed 及晋级门的机械判据 | rules/skills/drafts/workflow_landing_to_production.md |
| workflow_long_context_scale_up | 判断长上下文应预算分片并递归收口覆盖证明 | rules/skills/drafts/workflow_long_context_scale_up.md |
| workflow_manage_unexpected | 判断遇阻时应按动作阶梯逐级试、再精确 block | rules/skills/drafts/workflow_manage_unexpected.md |
| workflow_modular_design_entropy_control | 判断代码级模块化边界与补丁vs重构的度 | rules/skills/drafts/workflow_modular_design_entropy_control.md |
| workflow_phase_framework | 判断 Loop/AAU phase 计划是否良构与裁剪有据 | rules/skills/drafts/workflow_phase_framework.md |
| workflow_optimization_action_protocol | 判断优化循环中模型交替、兜底、审查与达成度收口 | rules/skills/drafts/workflow_optimization_action_protocol.md |
| workflow_long_multi_provider_run_protocol | 判断长多 provider 运行中降级、GOAL 忠实与检查回写 | rules/skills/drafts/workflow_long_multi_provider_run_protocol.md |
| workflow_orchestrator_mode | 判断主 Agent 应退化为纯编排者、派 worker 检测推进 | rules/skills/drafts/workflow_orchestrator_mode.md |
| workflow_post_loop_critical_review | 判断 Loop 完成后 Final Report 与完成声明是否可信 | rules/skills/drafts/workflow_post_loop_critical_review.md |
| workflow_precise_reading | 判断已定位信息是否按覆盖义务逐条读够 | rules/skills/drafts/workflow_precise_reading.md |
| workflow_recall_first_locate | 判断覆盖式收集时应先做机械召回工程 | rules/skills/drafts/workflow_recall_first_locate.md |
| workflow_recursive_agent_review | 判断 reviewer agent 是否真审了目标 agent | rules/skills/drafts/workflow_recursive_agent_review.md |
| workflow_change_visualization | 判断变更报告何时画、画哪种图、如何组织 | rules/skills/drafts/workflow_change_visualization.md |
| workflow_essence_diagramming | 判断触本质图示的图法路由与六条绘制纪律 | rules/skills/drafts/workflow_essence_diagramming.md |
| workflow_requirement_analysis_quality | 判断需求分析是否消息覆盖全且无偏置塌缩 | rules/skills/drafts/workflow_requirement_analysis_quality.md |
| workflow_requirement_decomposition_core | 判断任务定义面拆解与七门顺序骨架 | rules/skills/drafts/workflow_requirement_decomposition_core.md |
| workflow_requirement_intake | 判断新需求如何四文件入库与消费边定 phase | rules/skills/drafts/workflow_requirement_intake.md |
| workflow_session_claim_audit | 判断 session 完成声明是否如其声称 | rules/skills/drafts/workflow_session_claim_audit.md |
| workflow_solid_decision_phase | 判断多 closure 联动与疑问补料综合阶段 | rules/skills/drafts/workflow_solid_decision_phase.md |
| workflow_solid_decision_review | 判断证据等级、claim ceiling 与 next reaction | rules/skills/drafts/workflow_solid_decision_review.md |
| workflow_tool_skill_evolution | 判断工具/skill/方法论更新的演进流程与传播 | rules/skills/drafts/workflow_tool_skill_evolution.md |
| workflow_unit_decomposition_and_context_injection | 判断 Unit 分解边界、context 注入与熵管理 | rules/skills/drafts/workflow_unit_decomposition_and_context_injection.md |
| workflow_what_to_how_execution_bridge | 判断目标状态到首个真实切片的桥接路径 | rules/skills/drafts/workflow_what_to_how_execution_bridge.md |
| workflow_data_check_review | 判断 claimed-done 工作是否真对/真完的数据审查法 | rules/skills/drafts/workflow_data_check_review.md |
| workflow_file_storage_and_requirement_protocol | 判断 adhoc program 内部文件存放与需求记录的 SOP | rules/skills/drafts/workflow_file_storage_and_requirement_protocol.md |
| workflow_large_scale_task_audit | 判断大型多 session 任务是否真满足原始需求 | rules/skills/drafts/workflow_large_scale_task_audit.md |
| workflow_idea_to_production | 判断模糊想法如何推到真实条件下被用到 | rules/skills/drafts/workflow_idea_to_production.md |
| workflow_i2p_filter_compress | 判断海量调研材料如何筛出多候选方案 | rules/skills/drafts/workflow_i2p_filter_compress.md |
| workflow_module_integration_protocol | 判断模块如何以 MODULE.md 契约被注册与调用 | rules/skills/drafts/workflow_module_integration_protocol.md |
道-元路由 (4)
| 名称 | 一句话 | 真源路径 |
|---|---|---|
| dao_methodology_lookup | 判断动手前先查方法论库、按域选篇精读 | rules/skills/drafts/dao_methodology_lookup.md |
| dao_domain_competition | 判断竞赛任务先读优化段方法论与资产台账 | rules/skills/drafts/dao_domain_competition.md |
| workflow_report_writing_matrix | 判断写报告前如何按类型叠加内容与格式 skill | rules/skills/drafts/workflow_report_writing_matrix.md |
| workflow_guidance_skill_matrix | 判断道/术分类、联动与路由矩阵 | rules/skills/drafts/workflow_guidance_skill_matrix.md |
provenance：rules/skills/drafts/workflow_guidance_skill_matrix.md rules/skills/