{"id":"3928a337-b73b-495c-bbfa-b1fc52fb20cb","shortId":"DQHCUu","kind":"skill","title":"ai-native-pm-agent","tagline":"AI Native 产品经理 Agent — — 五书合一的产品工作流编排器。 用户提供一个产品想法或问题线索，Agent 自动编排七个阶段：","description":"# AI Native PM Agent\n\n## 使用场景\n\n- 你有一个产品想法，需要从需求发现到生产运行的完整产品方案\n- 你需要用系统化方法论管理 AI 产品全生命周期\n- 你希望五本书的方法论能在一个 Agent 中自然流转\n\n## 工作流\n\n```text\n用户输入\n  → [P1] 需求发现 (ai-native-product-needs/*)\n    → [P2] 方向定界 (ai-native-direction-framing)\n      → [P3] 体验设计 (ai-native-user-experience)\n        → [P4] 系统构建 (ai-native-system-building + 子 Skills)\n          → [P5] 商业模式 (ai-native-business-model/*)\n            → [P6] 增长策略 (ai-native-marketing-growth/*)\n              → [P7] 审计投产 (ai-native-audit-release  / 待创建)\n                → 输出完整产品方案\n```\n\n## 阶段详情\n\n### P1: 需求发现\n- **Skill**: `ai-native-product-needs/*`\n- **输入**: 用户原始想法\n- **输出**: 需求简报 (Needs Brief)\n- **关键判断**: 机会评分 >= 60 才进入 P2\n\n### P2: 方向定界\n- **Skill**: `ai-native-direction-framing`\n- **输入**: P1 需求简报\n- **输出**: Direction Brief\n- **关键判断**: go/no-go 决策\n\n### P3: 体验设计\n- **Skill**: `ai-native-user-experience`\n- **输入**: P2 Direction Brief\n- **输出**: UX 方案\n- **关键判断**: 人机协作模式定义\n\n### P4: 系统构建\n- **Skill**: `ai-native-system-building` + 子 Skills\n- **输入**: P3 UX 方案\n- **输出**: 系统架构方案\n- **包含**: Agent/Skill/记忆/上下文/RAG\n\n### P5: 商业模式\n- **Skill**: `ai-native-business-model/*`\n- **输入**: P4 系统成本估算\n- **输出**: 定价策略 + 收入模型\n- **关键判断**: 成本-定价匹配\n\n### P6: 增长策略\n- **Skill**: `ai-native-marketing-growth/*`\n- **输入**: P5 商业模式\n- **输出**: 获客/激活/留存方案\n- **关键判断**: 增长目标-系统承载匹配\n\n### P7: 审计投产\n- **Skill**: `ai-native-audit-release`\n- **输入**: P4-P6 全部方案\n- **输出**: go/no-go 决策 + 监控方案\n\n## 核心机制\n\n### 上下文传递\n\n每个阶段的输出自动写入 Product Context 的对应字段，下一阶段启动时自动读取。\n\n```yaml\nproduct_context:\n  version: \"1.0\"\n  created_at: \"2026-04-27T11:00:00+08:00\"\n  status: \"active\"  # active / paused / rollback\n  current_stage: \"p1\"  # p1-p7\n  \n  # 各阶段输出物\n  artifacts:\n    p1_needs_brief: null       # ai-native-product-needs 产出\n    p2_direction_brief: null   # ai-native-direction-framing 产出\n    p3_ux_design_brief: null   # ai-native-user-experience 产出\n    p4_system_architecture: null  # ai-native-system-building 产出\n    p5_business_model_brief: null # ai-native-business-model 产出\n    p6_growth_strategy_brief: null # ai-native-marketing-growth 产出\n    p7_audit_decision: null    # ai-native-audit-release 产出\n  \n  # 回溯历史\n  history:\n    - stage: \"p1\"\n      version: 1\n      completed_at: \"2026-04-27T11:30:00+08:00\"\n      artifacts_key: \"p1_needs_brief\"\n  \n  # 冲突检测结果\n  conflict_checks:\n    p4_p5_cost_pricing: null\n    p3_p4_ux_latency: null\n    p6_p4_growth_capacity: null\n```\n\n### 冲突检测\n每完成两个阶段，自动运行冲突检测：\n- P4 + P5 → 成本-定价匹配\n- P3 + P4 → UX-系统延迟匹配\n- P6 + P4 → 增长-承载匹配\n\n### 用户交互点\n以下节点必须等待用户确认：\n1. P2 结束后（方向定界 go/no-go）\n2. P4 结束后（系统架构方案确认）\n3. P7 结束后（最终投产决策）\n\n### 回溯机制\n用户可在任意时刻说：\n- \"回到 P2 重新看方向\" → 保留 P3-P7 为 v1，重新执行 P2 生成 v2\n- \"P4 的系统成本太高了\" → 回溯到 P4，P5-P7 标记为待更新\n\n## 使用方式\n\n当用户提供一个产品想法时，自动执行：\n\n1. 初始化 Product Context\n2. 按阶段路由依次执行各阶段 Skill\n3. 每阶段完成后进行冲突检测\n4. 用户可在任意节点确认/修改/回溯\n5. 最终输出完整产品方案\n\n## 核心概念\n\n### 概念一：总编排器的本质——不是串行流水线，而是智能路由\n\nAI Native PM Agent 不是把七个阶段串起来跑一遍。它的核心能力是：\n1. **智能路由**：根据当前阶段的输出决定下一步走哪条路径\n2. **冲突检测**：自动发现跨阶段的矛盾\n3. **回溯管理**：支持在任意节点回溯修改\n4. **上下文传递**：确保每个阶段都能获取前序阶段的完整上下文\n\n```\n用户输入\n  → 初始化 Product Context\n    → P1 需求发现 → [go/no-go]\n      → P2 方向定界 → [用户确认]\n        → P3 体验设计\n          → P4 系统构建 → [用户确认] → [冲突检测: P4 vs P5]\n            → P5 商业模式 → [冲突检测: P5 vs P6]\n              → P6 增长策略 → [冲突检测: P4 vs P6]\n                → P7 审计投产 → [用户确认] → 输出完整方案\n```\n\n> 总编排器的价值不是\"帮你跑完七个阶段\"，而是\"在每个阶段告诉你该不该继续、有没有冲突\"。\n\n### 概念二：Product Context——贯穿七阶段的共享状态\n\nProduct Context 是总编排器的核心数据结构，记录了每个阶段的产出物和状态：\n\n```yaml\nproduct_context:\n  version: \"1.0\"\n  status: \"active\"  # active / paused / rollback\n  current_stage: \"p1\"\n  \n  artifacts:\n    p1_needs_brief: null       # 需求发现产出\n    p2_direction_brief: null   # 方向定界定产\n    p3_ux_design_brief: null   # 体验设计产出\n    p4_system_architecture: null  # 系统构建产出\n    p5_business_model_brief: null # 商业模式产出\n    p6_growth_strategy_brief: null # 增长策略产出\n    p7_audit_decision: null    # 审计投产产出\n```\n\n每个阶段启动时自动读取前序阶段的产出，完成后自动写入自己的产出。\n\n### 概念三：三个冲突检测点\n\n每完成两个阶段，自动运行冲突检测：\n\n| 冲突检测 | 检查内容 | 典型问题 |\n|----------|----------|----------|\n| P4 + P5 | 成本-定价匹配 | 系统成本太高，定价无法覆盖 |\n| P3 + P4 | UX-系统延迟匹配 | UX 设计要求实时响应，系统架构做不到 |\n| P6 + P4 | 增长-承载匹配 | 增长目标超出系统承载能力 |\n\n> 冲突检测不是\"发现问题\"，而是\"在投入更多资源之前发现矛盾\"。\n\n### 概念四：三个用户交互点\n\n以下节点必须等待用户确认，不能自动继续：\n\n| 交互点 | 确认内容 | 不确认则 |\n|--------|----------|----------|\n| P2 结束后 | 方向定界 go/no-go | 暂停或回溯到 P1 |\n| P4 结束后 | 系统架构方案确认 | 回溯到 P3 或 P4 |\n| P7 结束后 | 最终投产决策 | 暂停或回溯 |\n\n### 概念五回溯机制\n\n用户可在任意时刻说：\n- \"回到 P2 重新看方向\" → 保留 P3-P7 为 v1，重新执行 P2 生成 v2\n- \"P4 的系统成本太高了\" → 回溯到 P4，P5-P7 标记为待更新\n- \"P6 的增长策略和 P5 的商业模式不匹配\" → 回溯到 P5 重新设计\n\n回溯不是失败，而是迭代。\n\n## 分步执行\n\n### Step 1：初始化 Product Context\n\n**输入**：用户原始想法\n\n**处理**：\n1. 创建 Product Context 数据结构\n2. 记录原始输入\n3. 设置 current_stage = p1\n4. 初始化所有 artifacts 为 null\n\n**输出**：Product Context（已初始化）\n\n---\n\n### Step 2：P1 需求发现\n\n**输入**：用户原始想法\n\n**处理**：\n1. 调用 p0-product-needs 或 p0-needs-orchestrator\n2. 执行微需求检测、四层拆解、真需求验证\n3. 输出 Needs Brief\n4. 检查机会评分：≥ 60 才进入 P2\n\n**输出**：Needs Brief → 写入 Product Context\n\n---\n\n### Step 3：P2 方向定界\n\n**输入**：P1 Needs Brief\n\n**处理**：\n1. 调用 ai-native-direction-framing\n2. 执行五维度判断 + 资料审查 + 能力评估\n3. 输出 Direction Brief\n4. **等待用户确认 go/no-go**\n\n**输出**：Direction Brief → 写入 Product Context\n\n---\n\n### Step 4：P3 体验设计\n\n**输入**：P2 Direction Brief\n\n**处理**：\n1. 调用 ai-native-user-experience\n2. 执行任务分析、人机分工、状态可见性、纠偏机制、信任设计\n3. 输出 UX 方案\n\n**输出**：UX Design Brief → 写入 Product Context\n\n---\n\n### Step 5：P4 系统构建 + P5 商业模式\n\n**输入**：P3 UX 方案 + P4 系统成本估算\n\n**处理**：\n1. P4：调用 ai-native-system-building，输出系统架构方案\n2. **冲突检测 P4 + P5**：系统成本 vs 定价是否匹配\n3. P5：调用 ai-native-business-model，输出定价策略 + 收入模型\n4. **等待用户确认系统架构方案**\n\n**输出**：System Architecture + Business Model Brief → 写入 Product Context\n\n---\n\n### Step 6：P6 增长策略 + P7 审计投产\n\n**输入**：P5 商业模式 + P4 系统架构\n\n**处理**：\n1. **冲突检测 P6 + P4**：增长目标 vs 系统承载是否匹配\n2. P6：调用 ai-native-marketing-growth，输出增长策略\n3. P7：调用 ai-native-audit-release，执行六类审计\n4. **等待用户确认最终投产决策**\n\n**输出**：Growth Strategy + Audit Decision → 写入 Product Context\n\n---\n\n### Step 7：输出完整产品方案\n\n**输入**：Product Context（全部 artifacts）\n\n**处理**：\n1. 整合七个阶段的产出物\n2. 标注所有冲突检测结果\n3. 标注所有用户确认点的状态\n4. 输出完整产品方案文档\n\n**输出**：完整产品方案（含需求发现、方向定界、体验设计、系统构建、商业模式、增长策略、审计投产）\n\n### 示例 1：AI 客服产品的完整编排\n\n**场景描述**：用户说\"我想做一个 AI 客服产品，帮企业自动回复客户问题\"。\n\n**Step 1 初始化**：\n```yaml\nproduct_context:\n  current_stage: \"p1\"\n  artifacts:\n    p1_needs_brief: null\n    # ... 全部 null\n```\n\n**Step 2 P1 需求发现**：\n- 微需求检测：发现新人客服的\"不敢问\"微需求\n- 四层拆解：表达层→场景层→处境层→代价层\n- 真需求验证：5/5 通过\n- 机会评分：85/100 → 进入 P2\n- **产出**：Needs Brief（核心问题：新人知识辅助，而非自动回复）\n\n**Step 3 P2 方向定界**：\n- 五维度判断：全部通过\n- 资料审查：FAQ + 历史工单可用\n- 能力评估：AI 辅助回复可行\n- **用户确认**：✅ go\n- **产出**：Direction Brief（聚焦新人客服知识辅助场景）\n\n**Step 4 P3 体验设计**：\n- 人机分工：AI 生成候选回复，人工确认后发送\n- 状态可见：展示 AI 置信度和引用来源\n- 纠偏机制：快速编辑、重新生成、完全拒绝\n- **产出**：UX Design Brief\n\n**Step 5 P4 + P5 冲突检测**：\n- P4 系统架构：RAG + 历史工单检索 + 候选回复生成\n- **冲突检测 P4 + P5**：✅ 成本可控（单次调用 < $0.01）\n- P5 商业模式：SaaS 订阅 + 按坐席收费\n- **用户确认**：✅ 系统架构方案通过\n- **产出**：System Architecture + Business Model Brief\n\n**Step 6 P6 + P7**：\n- **冲突检测 P6 + P4**：✅ 增长目标（月增 50 坐席）在系统承载范围内\n- P6 增长策略：PLG（免费试用 → 团队邀请 → 付费转化）\n- P7 审计投产：go（条件放行）\n- **用户确认**：✅ 最终投产\n- **产出**：Growth Strategy + Audit Decision\n\n**最终输出**：完整产品方案，从需求发现到投产的七个阶段全部完成。\n\n---\n\n### 示例 2：回溯场景——P4 系统成本过高\n\n**场景描述**：在执行到 P5 时，发现 P4 的系统架构方案成本过高，无法匹配 P5 的定价策略。\n\n**冲突检测触发**：\n```yaml\nconflict_check:\n  p4_p5_cost_pricing:\n    status: \"CONFLICT\"\n    detail: \"P4 系统方案的月运行成本 $50,000，但 P5 定价策略的月收入预期 $30,000\"\n    recommendation: \"回溯到 P4 优化系统架构，或回溯到 P5 调整定价\"\n```\n\n**用户决策**：\"回到 P4 重新设计系统架构\"\n\n**回溯处理**：\n1. 保留 P1-P3 的产出不变\n2. P4 标记为 v2，重新执行\n3. P5-P7 标记为\"待更新\"\n4. P4 v2 输出后，重新运行冲突检测\n5. 如果通过，继续执行 P5-P7 v2\n\n**P4 v2 优化**：\n- 将 RAG 检索从实时改为批量预处理\n- 将 LLM 调用从 GPT-4 降级为 GPT-3.5-turbo\n- 月运行成本从 $50,000 降至 $15,000\n\n**重新冲突检测**：✅ 成本-定价匹配\n\n> 回溯不是失败，而是迭代。好的编排器让回溯的成本最低。","tags":["native","agent","product","skills","gmaxxxie","agent-skills","ai-agent","ai-native","ai-product","methodology","product-management","product-methodology"],"capabilities":["skill","source-gmaxxxie","skill-ai-native-pm-agent","topic-agent-skills","topic-ai-agent","topic-ai-native","topic-ai-product","topic-methodology","topic-product-management","topic-product-methodology","topic-product-thinking","topic-skills"],"categories":["ai-native-product-agent-skills"],"synonyms":[],"warnings":[],"endpointUrl":"https://skills.sh/gmaxxxie/ai-native-product-agent-skills/ai-native-pm-agent","protocol":"skill","transport":"skills-sh","auth":{"type":"none","details":{"cli":"npx skills add gmaxxxie/ai-native-product-agent-skills","source_repo":"https://github.com/gmaxxxie/ai-native-product-agent-skills","install_from":"skills.sh"}},"qualityScore":"0.478","qualityRationale":"deterministic score 0.48 from registry signals: · indexed on github topic:agent-skills · 56 github stars · SKILL.md body (7,523 chars)","verified":false,"liveness":"unknown","lastLivenessCheck":null,"agentReviews":{"count":0,"score_avg":null,"cost_usd_avg":null,"success_rate":null,"latency_p50_ms":null,"narrative_summary":null,"summary_updated_at":null},"enrichmentModel":"deterministic:skill-github:v1","enrichmentVersion":1,"enrichedAt":"2026-05-18T18:57:24.601Z","embedding":null,"createdAt":"2026-05-01T01:02:50.398Z","updatedAt":"2026-05-18T18:57:24.601Z","lastSeenAt":"2026-05-18T18:57:24.601Z","tsv":"'+08':232,334 '-04':227,329 '-27':228,330 '-3.5':1142 '-4':1139 '/rag':157 '0.01':1007 '00':230,231,233,333,335 '000':1082,1087,1146,1149 '1':325,379,418,444,648,655,683,722,755,792,841,885,903,913,1100 '1.0':223,508 '15':1148 '2':384,422,447,660,677,694,729,762,801,848,887,929,1054,1106 '2026':226,328 '3':388,425,450,662,698,714,733,768,808,857,889,955,1111 '30':332,1086 '4':427,453,667,702,737,747,818,866,891,973,1117 '5':431,780,993,1122 '5/5':942 '50':1030,1081,1145 '6':830,1022 '60':100,704 '7':877 '85/100':945 'activ':235,236,510,511 'agent':5,9,12,17,25,441 'agent/skill':154 'ai':2,6,14,22,33,40,47,54,63,70,77,88,107,124,141,162,180,199,252,262,273,283,294,305,315,438,725,758,796,812,852,861,904,909,964,977,982 'ai-native-audit-releas':76,198,314,860 'ai-native-business-model':62,161,293,811 'ai-native-direction-fram':39,106,261,724 'ai-native-marketing-growth':69,179,304,851 'ai-native-pm-ag':1 'ai-native-product-ne':32,87,251 'ai-native-system-build':53,140,282,795 'ai-native-user-experi':46,123,272,757 'architectur':280,536,822,1017 'artifact':246,336,517,669,883,921 'audit':79,201,311,317,552,863,871,1048 'brief':97,116,131,249,259,270,291,302,340,520,525,531,542,548,701,709,720,736,742,753,775,825,924,950,970,991,1020 'build':57,144,286,799 'busi':65,164,289,296,540,814,823,1018 'capac':357 'check':343,1071 'complet':326 'conflict':342,1070,1077 'context':216,221,421,459,498,501,506,651,658,674,712,745,778,828,875,881,917 'cost':346,1074 'creat':224 'current':239,514,664,918 'decis':312,553,872,1049 'design':269,530,774,990 'detail':1078 'direct':42,109,115,130,258,264,524,727,735,741,752,969 'experi':50,127,276,761 'faq':961 'frame':43,110,265,728 'go':967,1041 'go/no-go':118,209,383,462,600,739 'gpt':1138,1141 'growth':73,183,300,308,356,546,855,869,1046 'histori':321 'key':337 'latenc':352 'llm':1136 'market':72,182,307,854 'model':66,165,290,297,541,815,824,1019 'nativ':3,7,15,34,41,48,55,64,71,78,89,108,125,142,163,181,200,253,263,274,284,295,306,316,439,726,759,797,813,853,862 'need':36,91,96,248,255,339,519,688,692,700,708,719,923,949 'null':250,260,271,281,292,303,313,348,353,358,521,526,532,537,543,549,554,671,925,927 'orchestr':693 'p0':686,691 'p0-needs-orchestrator':690 'p0-product-needs':685 'p1':30,84,112,241,243,247,323,338,460,516,518,602,666,678,718,920,922,930,1103 'p1-p3':1102 'p1-p7':242 'p2':37,102,103,129,257,380,395,404,463,523,597,617,626,706,715,751,947,956 'p3':44,120,148,267,349,367,399,466,528,572,607,621,748,786,974,1104 'p3-p7':398,620 'p4':51,137,167,205,278,344,350,355,362,368,373,385,407,410,468,472,484,534,565,573,581,603,609,629,632,781,789,793,803,838,844,994,997,1003,1027,1056,1063,1072,1079,1090,1097,1107,1118,1129 'p4-p6':204 'p5':60,158,185,288,345,363,412,474,475,478,539,566,634,639,642,783,804,809,836,995,1004,1008,1060,1066,1073,1084,1093,1113,1126 'p5-p7':411,633,1112,1125 'p6':67,176,206,299,354,372,480,481,486,545,580,637,831,843,849,1023,1026,1033 'p7':74,195,244,310,389,400,413,487,551,610,622,635,833,858,1024,1039,1114,1127 'paus':237,512 'plg':1035 'pm':4,16,440 'price':347,1075 'product':35,90,215,220,254,420,458,497,500,505,650,657,673,687,711,744,777,827,874,880,916 'rag':999,1133 'recommend':1088 'releas':80,202,318,864 'rollback':238,513 'saa':1010 'skill':59,86,105,122,139,146,160,178,197,424 'skill-ai-native-pm-agent' 'source-gmaxxxie' 'stage':240,322,515,665,919 'status':234,509,1076 'step':647,676,713,746,779,829,876,912,928,954,972,992,1021 'strategi':301,547,870,1047 'system':56,143,279,285,535,798,821,1016 't11':229,331 'text':28 'topic-agent-skills' 'topic-ai-agent' 'topic-ai-native' 'topic-ai-product' 'topic-methodology' 'topic-product-management' 'topic-product-methodology' 'topic-product-thinking' 'topic-skills' 'turbo':1143 'user':49,126,275,760 'ux':133,149,268,351,370,529,575,577,770,773,787,989 'ux-系统延迟匹配':369,574 'v1':402,624 'v2':406,628,1109,1119,1128,1130 'version':222,324,507 'vs':473,479,485,806,846 'yaml':219,504,915,1069 '三个冲突检测点':559 '三个用户交互点':591 '上下文':156 '上下文传递':213,454 '下一阶段启动时自动读取':218 '不敢问':934 '不是串行流水线':436 '不是把七个阶段串起来跑一遍':442 '不确认则':596 '不能自动继续':593 '中自然流转':26 '为':401,623,670 '五书合一的产品工作流编排器':10 '五维度判断':958 '交互点':594 '产出':256,266,277,287,298,309,319,948,968,988,1015,1045 '产品全生命周期':23 '产品经理':8 '人工确认后发送':979 '人机分工':764,976 '人机协作模式定义':136 '从需求发现到投产的七个阶段全部完成':1052 '付费转化':1038 '代价层':940 '以下节点必须等待用户确认':378,592 '优化':1131 '优化系统架构':1091 '但':1083 '体验设计':45,121,467,749,897,975 '体验设计产出':533 '你希望五本书的方法论能在一个':24 '你有一个产品想法':19 '你需要用系统化方法论管理':21 '使用场景':18 '使用方式':415 '保留':397,619,1101 '信任设计':767 '修改':429 '候选回复生成':1001 '免费试用':1036 '全部':882,926 '全部方案':207 '全部通过':959 '关键判断':98,117,135,172,191 '典型问题':564 '写入':710,743,776,826,873 '冲突检测':359,448,471,477,483,562,802,842,996,1002,1025 '冲突检测不是':586 '冲突检测结果':341 '冲突检测触发':1068 '决策':119,210 '分步执行':646 '创建':656 '初始化':419,457,649,914 '初始化所有':668 '包含':153 '单次调用':1006 '历史工单可用':962 '历史工单检索':1000 '发现':1062 '发现新人客服的':933 '发现问题':587 '各阶段输出物':245 '含需求发现':895 '商业模式':61,159,186,476,784,837,899,1009 '商业模式产出':544 '四层拆解':696,936 '回到':394,616,1096 '回溯':430 '回溯不是失败':644,1154 '回溯到':409,606,631,641,1089 '回溯历史':320 '回溯场景':1055 '回溯处理':1099 '回溯机制':392 '回溯管理':451 '团队邀请':1037 '在执行到':1059 '在投入更多资源之前发现矛盾':589 '在每个阶段告诉你该不该继续':494 '在系统承载范围内':1032 '场景层':938 '场景描述':906,1058 '坐席':1031 '增长':375,583 '增长-承载匹配':374,582 '增长目标':193,845,1028 '增长目标-系统承载匹配':192 '增长目标超出系统承载能力':585 '增长策略':68,177,482,832,900,1034 '增长策略产出':550 '处境层':939 '处理':654,682,721,754,791,840,884 '好的编排器让回溯的成本最低':1156 '如果通过':1123 '子':58,145 '它的核心能力是':443 '完全拒绝':987 '完成后自动写入自己的产出':557 '完整产品方案':894,1051 '定价匹配':175,366,569,1153 '定价无法覆盖':571 '定价是否匹配':807 '定价策略':170 '定价策略的月收入预期':1085 '审计投产':75,196,488,834,901,1040 '审计投产产出':555 '客服产品':910 '客服产品的完整编排':905 '将':1132,1135 '展示':981 '工作流':27 '已初始化':675 '帮企业自动回复客户问题':911 '帮你跑完七个阶段':492 '当用户提供一个产品想法时':416 '待创建':81 '待更新':1116 '微需求':935 '微需求检测':932 '快速编辑':985 '总编排器的价值不是':491 '总编排器的本质':435 '成本':174,365,568,1152 '成本-定价匹配':173,364,567,1151 '成本可控':1005 '我想做一个':908 '或':608,689 '或回溯到':1092 '才进入':101,705 '执行五维度判断':730 '执行任务分析':763 '执行六类审计':865 '执行微需求检测':695 '承载匹配':376,584 '按坐席收费':1012 '按阶段路由依次执行各阶段':423 '支持在任意节点回溯修改':452 '收入模型':171,817 '数据结构':659 '整合七个阶段的产出物':886 '新人知识辅助':952 '方向定界':38,104,382,464,599,716,896,957 '方向定界定产':527 '方案':134,150,771,788 '无法匹配':1065 '时':1061 '是总编排器的核心数据结构':502 '智能路由':445 '暂停或回溯':613 '暂停或回溯到':601 '最终投产':1044 '最终投产决策':391,612 '最终输出':1050 '最终输出完整产品方案':432 '月增':1029 '月运行成本从':1144 '有没有冲突':495 '机会评分':99,944 '条件放行':1042 '标注所有冲突检测结果':888 '标注所有用户确认点的状态':890 '标记为':1108,1115 '标记为待更新':414,636 '核心机制':212 '核心概念':433 '核心问题':951 '根据当前阶段的输出决定下一步走哪条路径':446 '检查内容':563 '检查机会评分':703 '检索从实时改为批量预处理':1134 '概念一':434 '概念三':558 '概念二':496 '概念五回溯机制':614 '概念四':590 '每个阶段启动时自动读取前序阶段的产出':556 '每个阶段的输出自动写入':214 '每完成两个阶段':360,560 '每阶段完成后进行冲突检测':426 '激活':189 '状态可见':980 '状态可见性':765 '生成':405,627 '生成候选回复':978 '用户交互点':377 '用户决策':1095 '用户原始想法':93,653,681 '用户可在任意时刻说':393,615 '用户可在任意节点确认':428 '用户提供一个产品想法或问题线索':11 '用户确认':465,470,489,966,1013,1043 '用户说':907 '用户输入':29,456 '留存方案':190 '的产出不变':1105 '的商业模式不匹配':640 '的增长策略和':638 '的定价策略':1067 '的对应字段':217 '的系统成本太高了':408,630 '的系统架构方案成本过高':1064 '监控方案':211 '真需求验证':697,941 '确保每个阶段都能获取前序阶段的完整上下文':455 '确认内容':595 '示例':902,1053 '等待用户确认':738 '等待用户确认最终投产决策':867 '等待用户确认系统架构方案':819 '系统延迟匹配':371,576 '系统成本':805 '系统成本估算':168,790 '系统成本太高':570 '系统成本过高':1057 '系统承载匹配':194 '系统承载是否匹配':847 '系统方案的月运行成本':1080 '系统构建':52,138,469,782,898 '系统构建产出':538 '系统架构':839,998 '系统架构做不到':579 '系统架构方案':152 '系统架构方案确认':387,605 '系统架构方案通过':1014 '纠偏机制':766,984 '结束后':381,386,390,598,604,611 '继续执行':1124 '置信度和引用来源':983 '而是':493,588 '而是智能路由':437 '而是迭代':645,1155 '而非自动回复':953 '聚焦新人客服知识辅助场景':971 '能力评估':732,963 '自动发现跨阶段的矛盾':449 '自动执行':417 '自动编排七个阶段':13 '自动运行冲突检测':361,561 '获客':188 '表达层':937 '订阅':1011 '记录了每个阶段的产出物和状态':503 '记录原始输入':661 '记忆':155 '设置':663 '设计要求实时响应':578 '调整定价':1094 '调用':684,723,756,794,810,850,859 '调用从':1137 '贯穿七阶段的共享状态':499 '资料审查':731,960 '辅助回复可行':965 '输入':92,111,128,147,166,184,203,652,680,717,750,785,835,879 '输出':94,114,132,151,169,187,208,672,699,707,734,740,769,772,820,868,893 '输出后':1120 '输出增长策略':856 '输出完整产品方案':82,878 '输出完整产品方案文档':892 '输出完整方案':490 '输出定价策略':816 '输出系统架构方案':800 '进入':946 '通过':943 '重新冲突检测':1150 '重新执行':403,625,1110 '重新生成':986 '重新看方向':396,618 '重新设计':643 '重新设计系统架构':1098 '重新运行冲突检测':1121 '阶段详情':83 '降级为':1140 '降至':1147 '需求发现':31,85,461,679,931 '需求发现产出':522 '需求简报':95,113 '需要从需求发现到生产运行的完整产品方案':20","prices":[{"id":"48a13c90-b17b-4d58-a9b0-a9d21e959318","listingId":"3928a337-b73b-495c-bbfa-b1fc52fb20cb","amountUsd":"0","unit":"free","nativeCurrency":null,"nativeAmount":null,"chain":null,"payTo":null,"paymentMethod":"skill-free","isPrimary":true,"details":{"org":"gmaxxxie","category":"ai-native-product-agent-skills","install_from":"skills.sh"},"createdAt":"2026-05-01T01:02:50.398Z"}],"sources":[{"listingId":"3928a337-b73b-495c-bbfa-b1fc52fb20cb","source":"github","sourceId":"gmaxxxie/ai-native-product-agent-skills/ai-native-pm-agent","sourceUrl":"https://github.com/gmaxxxie/ai-native-product-agent-skills/tree/main/skills/ai-native-pm-agent","isPrimary":false,"firstSeenAt":"2026-05-01T01:02:50.398Z","lastSeenAt":"2026-05-18T18:57:24.601Z"}],"details":{"listingId":"3928a337-b73b-495c-bbfa-b1fc52fb20cb","quickStartSnippet":null,"exampleRequest":null,"exampleResponse":null,"schema":null,"openapiUrl":null,"agentsTxtUrl":null,"citations":[],"useCases":[],"bestFor":[],"notFor":[],"kindDetails":{"org":"gmaxxxie","slug":"ai-native-pm-agent","github":{"repo":"gmaxxxie/ai-native-product-agent-skills","stars":56,"topics":["agent-skills","ai-agent","ai-native","ai-product","methodology","product-management","product-methodology","product-thinking","skills"],"license":null,"html_url":"https://github.com/gmaxxxie/ai-native-product-agent-skills","pushed_at":"2026-05-06T07:19:36Z","description":"AI Native Product Methodology — 80 executable skills across P0-P14 stages, covering needs discovery to aesthetic authority. From 8 books.","skill_md_sha":"18b5b3e65ddc967c6f2c40082bdc717524dc25dc","skill_md_path":"skills/ai-native-pm-agent/SKILL.md","default_branch":"main","skill_tree_url":"https://github.com/gmaxxxie/ai-native-product-agent-skills/tree/main/skills/ai-native-pm-agent"},"layout":"multi","source":"github","category":"ai-native-product-agent-skills","frontmatter":{"name":"ai-native-pm-agent","description":"AI Native 产品经理 Agent — — 五书合一的产品工作流编排器。 用户提供一个产品想法或问题线索，Agent 自动编排七个阶段："},"skills_sh_url":"https://skills.sh/gmaxxxie/ai-native-product-agent-skills/ai-native-pm-agent"},"updatedAt":"2026-05-18T18:57:24.601Z"}}