{"meta":{"title":"我的成长记录","lastUpdated":"2026-09-03T13:58:13+08:00","dataFlow":"个人目标与参考材料综合为领域、模块和具体学习内容；每日资讯独立扫描真实新变化 → 价值准入 → 本人判断 → 确认后沉淀到博客","sourcePolicy":"成熟课程与行业标准提供稳定骨架；每日资讯只负责修订资源、重要性和实践方式，或提出经过筛选的新候选叶子。博客文章永不作为输入。","privacy":"不公开本人选择、自评、打卡和详细复盘"},"masteredCapabilities":[],"informationDigests":[],"skills":{"summary":{"personal":5},"items":[{"id":"linn-growth-coach","kind":"personal","name":"AI 成长教练","icon":"◇","category":"个人成长系统","maturity":"试运行","overview":"以真实项目和工作事项为主线维护每日、每周复盘，并把已确认的可复用经验组织为统一的能力资产。","whenToUse":["每日 / 每周复盘","项目成果与能力证据","AI 系统质量工程"],"inputs":["成长档案","本人选择与自评","真实工作反馈","外部一手技术与岗位信号"],"workflow":["确认输入与边界","按 Skill 契约执行","验证并交付结果"],"outputs":["项目化每日与每周复盘","工作小步与当前进度","经验、踩坑与下一步","能力资产更新建议","Blog 安全摘要"],"limitations":["待连续运行两周后判断是否提升为已验证。"],"lastUsed":"2026-09-03"},{"id":"jmeter-business-chain-builder","kind":"personal","name":"JMeter 业务链压测构建器","icon":"◇","category":"性能测试工程化","maturity":"已验证","overview":"把抓包记录沉淀为可验证、可复用的 JMeter 业务链压测脚本和交互式报告。","whenToUse":["自动化测试架构","全链路性能瓶颈分析","项目成果与能力证据"],"inputs":["HTTP 抓包","参考 JMX","业务链约束"],"workflow":["确认输入与边界","按 Skill 契约执行","验证并交付结果"],"outputs":["完整 JMX","单一 JTL","用户与接口下钻报告","校验结果"],"limitations":["公开展示必须隐藏公司、账号、接口与环境信息。"],"lastUsed":"2026-07-23"},{"id":"publish-growth-workspace","kind":"personal","name":"成长工作台发布器","icon":"◇","category":"知识与发布系统","maturity":"试运行","overview":"同步私人工作台与 Blog 的能力资产投影，保持单一能力入口、隐私边界、持久存储和发布结果一致。","whenToUse":["知识系统","隐私投影","Sites 发布"],"inputs":["工作台私有目录","确认后的公开摘要","Sites 工作台","Blog 发布页面"],"workflow":["确认输入与边界","按 Skill 契约执行","验证并交付结果"],"outputs":["可展开的能力资产详情","隐私安全的公开能力投影","私人工作台","双平台发布验证"],"limitations":["首次验证中；私人自评和详细复盘不得自动公开。"],"lastUsed":"2026-08-17"},{"id":"ai-assisted-tracking-validation","kind":"personal","name":"AI 埋点验收助手","icon":"◇","category":"AI 测试与数据质量","maturity":"已验证","overview":"把埋点文档、真机或浏览器操作、App/运行时日志、SDK 或网络接收证据、分析控制台与验收表串成可追溯的验证流程。","whenToUse":["AI 系统质量工程","客户端数据质量","自动化测试架构","项目成果与能力证据"],"inputs":["埋点文档","已授权测试设备","测试账号与环境","Analytics 调试控制台或 Web 网络证据"],"workflow":["确认输入与边界","按 Skill 契约执行","验证并交付结果"],"outputs":["逐事件验收矩阵","验收运行档案","证据索引与分层原始证据","失败问题汇总","Excel/CSV 验收报告"],"limitations":["需要真实设备或可操作测试环境才能给出运行结论","验证码、敏感权限和第三方授权仍需本人控制","跨项目或更换分析平台时，需要重新核对事件与平台规则"],"lastUsed":"2026-08-21"},{"id":"linn-technical-blog-writer","kind":"personal","name":"Linn 技术博文写作助手","icon":"◇","category":"知识与发布系统","maturity":"试运行","overview":"把已经完成的真实技术工作整理为有证据、可脱敏、分类一致且符合 Linn-sBlog 规范的中文 Hexo 技术博文。","whenToUse":["项目成果与能力证据","技术写作","知识系统"],"inputs":["真实工作证据","复盘要点","当前官方资料","Linn-sBlog 格式约定","现有文章分类体系"],"workflow":["确认输入与边界","按 Skill 契约执行","验证并交付结果"],"outputs":["Hexo Markdown 博文","脱敏技术叙事","封面元数据","本地构建验证结果"],"limitations":["已完成第二次真实纠错使用，成熟度仍保持试运行；待非测试类文章验证结构与分类泛化。"],"lastUsed":"2026-08-12"}]},"capabilityPackages":{"updatedAt":"2026-08-20T19:54:21+08:00","items":[{"id":"ai-tracking-validation","name":"AI 辅助埋点验收","domain":"数据质量","summary":"用 AI 串联埋点规范、真机操作、客户端日志、分析平台核验与验收报告，降低逐项人工核对和问题定位成本。","problem":"解决埋点验收依赖人工逐项核对、证据分散和失败原因难以定位的问题。","usageExample":{"title":"例如：验收一次注册流程的移动端埋点","description":"测试人员在手机上完成注册，AI 同时对齐埋点规范、App 日志、Analytics SDK 诊断和 DebugView，最后给出每个事件是否到达、参数是否正确以及失败发生在哪一层。"},"whenToUse":["需要验收事件名、触发时机、参数、用户属性或事件顺序时","App 日志显示已上报，但分析平台没有收到事件时","需要形成 Android、iOS 或 Web 的可追溯埋点验收报告时"],"inputs":["埋点规范或验收表","已授权测试设备与测试包","测试环境和账号","客户端、运行时、SDK 或 Web 网络日志","DebugView 或等价调试控制台"],"skillIds":["ai-assisted-tracking-validation"],"workflow":[{"name":"解析规范","actor":"AI","action":"整理事件、参数、用户属性和触发条件。","output":"验收矩阵"},{"name":"静态预检","actor":"自动","action":"检查事件名、参数约束和 SDK 拒收风险。","output":"预检问题"},{"name":"触发场景","actor":"AI + 工具","action":"在已授权设备执行短业务流程。","output":"执行记录"},{"name":"关联证据","actor":"AI + 工具","action":"对齐客户端日志、运行时异常、SDK/网络接收和调试控制台。","output":"逐事件证据链"},{"name":"判定交付","actor":"AI，人工确认","action":"仅在目标事件获得接收层证据后判通过，并分类缺失、拒收、未执行、阻塞和重复。","output":"验收报告"}],"tools":[{"name":"Firebase 埋点规范预检脚本","kind":"脚本","status":"已存在","role":"在真机执行前检查事件与参数命名、参数数量、缺失触发条件及跨平台参数结构差异。","source":"ai-assisted-tracking-validation/scripts/lint_tracking_spec.py + lint_firebase_names.py"},{"name":"本地验收运行工具组","kind":"脚本","status":"已存在","role":"创建唯一运行目录，记录采集窗口、带哈希证据和结论修改历史，并在证据层级、状态和总数一致后生成权威汇总。","source":"ai-assisted-tracking-validation/scripts/start_validation_run.py + set_validation_capture.py + add_validation_evidence.py + set_validation_result.py + finalize_validation_run.py"},{"name":"Web 埋点载荷安全预检","kind":"脚本","status":"已存在","role":"检查站外 H5/API JSON 中要求为字符串的字段、数值型 ID 和 JavaScript 非安全整数，降低运行时类型与精度问题进入埋点和归因链路的风险。","source":"ai-assisted-tracking-validation/scripts/lint_web_tracking_payload.py"},{"name":"ADB / iOS 设备日志工具","kind":"外部工具","status":"已使用","role":"读取 App 埋点调用和 Analytics SDK 的接收、拒收日志。","source":"Android adb logcat；iOS 设备控制台与调试日志"},{"name":"Firebase DebugView","kind":"外部平台","status":"已使用","role":"核对事件是否到达分析平台，以及参数、顺序和用户属性。","source":"Firebase Analytics DebugView"},{"name":"埋点验收团队运行模块","kind":"项目工具","status":"试运行","role":"提供平台任务 API/UI、持久化状态、结构化导入、可配置 Planner、审批、Runner 队列、证据、判定、取消、恢复、租约看门狗、验收和审计。","source":"测试平台/backend/app/api/v1/tracking_validation.py + frontend/src/views/tracking"},{"name":"Android 埋点 Runner 与动作 Adapter","kind":"项目工具","status":"待真实环境验收","role":"建立 Android Logcat 采集窗口并上传逐事件证据；自动模式只执行埋点触发所需的白名单动作，不提供 UI 用例或 UI 断言。","source":"测试平台/tracking-runner"}],"harnesses":[{"name":"平台埋点验收主 Harness","status":"试运行","role":"从原始规范开始，编排 Planner、人工计划 Gate、Android Runner、运行状态、证据、Verifier、停止、取消、恢复和审计；它是位于 Skill 与 Adapter 之上的能力包级控制器。","includes":["数据库状态、过程时间线与任务 checkpoint","结构化导入、可配置 AI Planner 与人工计划门禁","Runner 排队、停止、取消、断点恢复与心跳租约看门狗","证据 SHA-256 完整性和接收层 Pass 门禁","脱敏事件轨迹、逐事件结论与最终 Verifier"],"source":"测试平台/backend/app/services/tracking_validation.py + backend/app/api/v1/tracking_validation.py"},{"name":"离线/契约 TrackingValidationHarness","status":"试运行","role":"用 JSON checkpoint 独立验证同一套状态、Policy、Adapter 接口、停止、恢复和验收契约；它不是第二个能力包，也不是平台主入口。","includes":["原子 JSON checkpoint 与只追加脱敏审计","Mock Adapter、取消、恢复和 watchdog","证据完整性与最终 Verifier"],"source":"测试平台/tools/tracking_validation/harness.py"}],"harnessNote":"能力包级平台 Harness、API/UI、可配置 Planner 和 Android Runner 已实现并通过软件验证；真实 Provider、Android 真机/Appium 链仍待环境验收，iOS/Web/Console Adapter、独立定时 Worker、真实 AI Eval 和数据飞轮尚未建设。","evaluations":[{"name":"Planner 输出评分接口","status":"仅基础设施","role":"独立计算用例召回、事件准确率、参数召回和幻觉率；尚无 held-out 数据集与真实模型运行成绩。","source":"测试平台/tools/tracking_validation/evals/grader.py"}],"crossCutting":[{"name":"Observability / Audit","role":"run_id、原子 checkpoint、只追加脱敏轨迹、采集起止时间、证据哈希、逐事件引用和结论修改历史形成可追溯证据链。"},{"name":"Security","role":"账号、内部地址、日志敏感字段和令牌不进入公开报告。"}],"contracts":[{"name":"埋点验收矩阵","role":"统一预期、实际结果、证据和最终结论。"},{"name":"验收运行档案与证据索引","role":"用同一 run_id 固定规范版本、环境、构建、证据文件哈希和逐事件引用。"},{"name":"结果分类","role":"统一通过、失败、未执行、阻塞、不适用和待确认口径；代码存在、UI 已执行和自动页面事件不能替代目标事件接收证据。"},{"name":"结论修改历史与权威汇总","role":"保留每次判定变化，并用验收矩阵、证据索引与自动生成汇总的哈希约束最终交付一致性。"},{"name":"团队规范化任务与学习记录 Schema","role":"统一 Planner、Harness、Adapter、Verifier 的任务字段，并为经人工批准的脱敏学习候选记录数据集版本和来源血缘。"}],"outputs":["逐事件验收矩阵","本地验收运行目录","证据索引与分层证据","完成性汇总","Excel / CSV 验收报告"],"boundaries":["需要真实设备或可操作测试环境。","验证码、敏感权限和第三方授权由 Linn 控制。","公司事件、账号、日志和内部环境不进入公开版本。"],"evolution":"从单次 Android/iOS/Web 验收方法，迭代为带证据门禁和结论历史的个人 Skill；08-20 建成离线契约 Harness，08-21 接入平台 API/UI、持久化主 Harness、可配置 Planner 与 Android Runner。当前先完成真实 Provider 和 Android 真机/Appium 试运行，再建设其他 Adapter、held-out Eval 和数据闭环。"},{"id":"jmeter-business-chain","name":"JMeter 业务链性能测试","domain":"性能测试","summary":"将抓包、业务链建模、JMeter 脚本校验、单次执行和结果下钻连接起来，减少复杂压测脚本的重复搭建成本。","problem":"减少复杂业务链脚本的重复搭建、参数拼接和结果分析成本。","usageExample":{"title":"例如：把一组真实抓包转换为可重复执行的登录后业务链","description":"从客户端抓包识别登录、后台轮询和业务分支，生成完整 JMX；平台执行器负责启动、监控和取消 JMeter，单一 JTL 保存权威结果，报告按用户和接口下钻。"},"whenToUse":["需要把真实 HTTP 抓包转成可维护的 JMeter 业务链时","业务链包含登录复用、后台流量、随机完整分支或动态关联时","需要从一次 JTL 按用户、业务链和接口分析结果时"],"inputs":["按真实客户端顺序采集的 HTTP 请求","参考 JMX 与登录链","账号 CSV 和运行参数","业务分支、节奏与验收约束"],"skillIds":["jmeter-business-chain-builder"],"workflow":[{"name":"读取输入","actor":"AI","action":"识别抓包、登录复用、后台流量和业务分支。","output":"业务链模型"},{"name":"生成 JMX","actor":"AI + 工具","action":"组装线程组、控制器、参数和关联规则。","output":"完整 JMX"},{"name":"静态校验","actor":"自动","action":"检查 XML、变量引用、断言和插件依赖。","output":"校验结果"},{"name":"验证执行","actor":"自动","action":"运行最小验证并写入单一结果源。","output":"单一 JTL"},{"name":"结果下钻","actor":"自动","action":"按用户、接口和业务分支组织结果。","output":"交互报告"}],"tools":[{"name":"Apache JMeter 5.6.3","kind":"外部软件","status":"已使用","role":"加载和执行 JMX，并把一次运行写入单一 JTL。","source":"Apache JMeter CLI"},{"name":"压测执行器","kind":"项目工具","status":"已存在","role":"由测试平台启动 JMeter、传入运行参数并管理结果。","source":"测试平台/jmeter-executor"},{"name":"交互报告生成器","kind":"项目脚本","status":"已存在","role":"把单一 JTL 转换为用户、业务链与接口下钻报告。","source":"generate_stress_report.sh + report_dimensions.py"}],"harnesses":[{"name":"JMeter 运行与结果子 Harness","status":"已形成","role":"管理已经生成的 JMX 从启动、进程监控、取消和结果收集到报告生成的执行生命周期；抓包理解和 JMX 设计仍由 Skill 与工具协作完成，不归入该 Harness。","includes":["JMeter CLI 启动与运行参数注入","进程取消、看门狗与 JTL 大小保护","单一 JTL 权威结果源","结果解析、压缩与回调","用户、业务链和接口下钻报告"],"source":"测试平台/jmeter-executor + generate_stress_report.sh + report_dimensions.py"}],"harnessNote":"已有真实运行子 Harness，但范围仅是已有 JMX 的运行与结果生命周期；尚无能力包级 Agent Harness 掌控抓包建模、JMX 生成、Gate、运行和验收全过程。","evaluations":[],"crossCutting":[{"name":"Observability / Audit","role":"单一 JTL、执行日志和可下钻报告保存一次运行的接口与用户证据。"},{"name":"Security","role":"交付前扫描 Token、设备标识、账号、内部接口和其他抓包秘密。"}],"contracts":[{"name":"业务链配置","role":"描述冷启动登录、后台流量和随机业务分支。"},{"name":"单一 JTL 约定","role":"确保一次执行只有一个权威结果源。"}],"outputs":["可执行 JMX","单一 JTL","用户与接口下钻报告","脚本校验结果"],"boundaries":["真实账号、接口、环境和业务参数只留在私有执行配置。","结构校验通过不等于目标系统已经完成容量验证。"],"evolution":"从脚本生成扩展到业务链校验、单一结果文件和用户/接口下钻报告。"},{"id":"personal-growth-publishing","name":"AI 驱动的个人成长与知识发布系统","domain":"知识与发布","summary":"用私人工作台承接真实工作、复盘和能力资产，再把本人确认且完成脱敏的内容投影到公开博客。","problem":"解决工作记录、能力沉淀和公开展示彼此割裂，以及私人信息容易误公开的问题。","usageExample":{"title":"例如：把一项真实测试实践沉淀为私人能力，再选择是否公开","description":"先在私人工作台记录真实进度、Skill、Tool 和验证证据；只有完成、确认且脱敏的部分才投影到 Blog。私人复盘、账号、内部系统和未完成内容始终留在私有侧。"},"whenToUse":["需要把真实工作、复盘、Skill 和项目产物整理为统一能力资产时","需要从私人工作台生成隐私安全的 Blog 能力投影时","需要同步、构建和验证 Growth Lab 与 Blog 数据契约时"],"inputs":["私人工作台权威数据","真实工作与资产证据","本人确认的公开范围","Growth Lab 和 Blog 构建环境"],"skillIds":["linn-growth-coach","publish-growth-workspace","linn-technical-blog-writer"],"workflow":[{"name":"记录工作","actor":"人工 + 系统","action":"保存随手记、任务、项目和实际进展。","output":"私人上下文"},{"name":"工作复盘","actor":"AI + 人工","action":"按项目整理草稿，确认后留档。","output":"确认复盘"},{"name":"关联能力","actor":"AI","action":"把真实形成的资产挂到对应能力包。","output":"能力目录"},{"name":"公开投影","actor":"AI，人工确认","action":"只选择已完成并脱敏的能力说明。","output":"公开数据"},{"name":"构建发布","actor":"自动 + AI","action":"验证页面、持久化和隐私边界。","output":"Growth Lab 与 Blog"}],"tools":[{"name":"能力目录构建脚本","kind":"项目脚本","status":"已存在","role":"从私人源数据生成工作台目录和脱敏后的 Blog 投影。","source":"apps/growth-lab/scripts/build-growth-catalog.mjs"},{"name":"工作台目录同步脚本","kind":"项目脚本","status":"已存在","role":"把构建后的私人目录同步到受保护的 Growth Lab 数据源。","source":"apps/growth-lab/scripts/sync-workspace-content.mjs"},{"name":"Growth Workspace Release Harness","kind":"项目工具","status":"试运行","role":"按检查点串联目录生成、代码检查、测试、Blog 生成、远程同步和部署，并明确区分本地完成与已部署；线上验收仍需单独核对。","source":"apps/growth-lab/scripts/growth-release-harness.mjs"},{"name":"Growth Lab 自动测试","kind":"测试套件","status":"已存在","role":"验证认证、公开投影、服务端渲染、数据契约与隐私字段。","source":"apps/growth-lab/tests/*.test.mjs"},{"name":"Hexo","kind":"外部软件","status":"已使用","role":"生成并发布公开博客静态页面。","source":"Hexo generate / deploy"}],"harnesses":[{"name":"成长工作区发布子 Harness","status":"试运行","role":"管理私人目录和公开投影从本地构建、验证、授权停止点、远程同步到部署的发布生命周期。","includes":["私有目录与公开投影构建","Lint、自动测试、Growth Lab 构建和 Hexo 生成","0600 原子 checkpoint 与已完成阶段跳过","远程发布显式授权门禁","工作台同步回读、Worker 本地认证冒烟与部署","Blog 部署和失败阶段记录"],"source":"apps/growth-lab/scripts/growth-release-harness.mjs"}],"harnessNote":"已形成可运行的发布子 Harness。默认只完成本地验证并记录 remotePending；授权后可续跑同步与部署，但它不控制前置复盘、能力判断和公开选择，也不替代单独的线上验收。","evaluations":[],"crossCutting":[{"name":"Privacy / Security","role":"私人权威源单向生成公开投影，公开侧不反向输入，未批准内容默认私有。"},{"name":"Audit","role":"activity.log、协作记录、D1 修订和发布版本记录关键变更。"},{"name":"Recovery","role":"D1 导出恢复、修订历史、Git 源码和发布 Harness checkpoint 提供实际恢复路径，但不宣称绝对不会丢失。"}],"contracts":[{"name":"单向发布契约","role":"工作台是权威源，Blog 不反向决定方向、等级或证据。"},{"name":"复盘确认契约","role":"Codex 先展示草稿，Linn 确认后才留档。"},{"name":"能力公开契约","role":"私人可跟踪未完成能力，公开只显示已形成内容。"}],"outputs":["私人工作台","工作复盘与修订记录","能力资产目录","脱敏后的公开能力页面"],"boundaries":["私人选择、自评、复盘、公司信息和未批准内容不进入 Blog。","不会为了展示而制造 Demo、证据、评测或技术文档。","第三方托管存在平台可用性边界。"],"evolution":"从平铺记录和 Skill 列表，逐步形成工作台管理过程、博客展示成熟能力的单向发布结构。"}]}}