marketing-content
Compare original and translation side by side
🇺🇸
Original
English🇨🇳
Translation
ChineseMarketing Content Skill
营销内容技能
Overview
概述
Enable agents to produce high-quality, data-driven marketing content using proven frameworks, structured workflows, and measurable performance loops.
marketing-strategistThis skill covers the full content lifecycle:
- Strategy → audience mapping, funnel alignment, channel selection
- Creation → copywriting frameworks, platform adaptation
- Operations → editorial calendar, scheduling, workflow
- Optimization → A/B testing, iteration, performance KPIs
- Measurement → engagement, conversion, retention, ROI
让 Agent能够使用经过验证的框架、结构化工作流和可衡量的效果闭环,生成高质量、数据驱动的营销内容。
marketing-strategist本技能覆盖完整的内容生命周期:
- 策略 → 受众画像映射、漏斗匹配、渠道选择
- 创作 → 文案写作框架、平台适配
- 运营 → 编辑日历、排期、工作流
- 优化 → A/B测试、迭代、效果KPIs
- 衡量 → engagement、转化率、留存率、ROI
When to Use
使用场景
Invoke when asked to:
- Write or plan marketing copy (ads, emails, blog posts, social)
- Build or maintain an editorial calendar
- Design a content campaign with objectives and KPIs
- Optimize existing content performance
- Conduct audience segmentation for content targeting
- A/B test content variants
- Adapt content for a specific platform (LinkedIn, TikTok, Email, etc.)
当收到以下需求时调用:
- 撰写或策划营销文案(广告、邮件、博客文章、社交内容)
- 搭建或维护编辑日历
- 设计带目标和KPI的内容活动
- 优化现有内容表现
- 为内容定位做受众分群
- 对内容变体做A/B测试
- 为特定平台适配内容(LinkedIn、TikTok、邮件等)
Enforcement Hooks
执行钩子
Input validated against before execution.
Output contract defined in .
schemas/input.schema.jsonschemas/output.schema.jsonPre-execution hook:
Post-execution hook:
hooks/pre-execute.cjshooks/post-execute.cjs执行前输入会通过校验。
输出规范定义在中。
schemas/input.schema.jsonschemas/output.schema.json执行前钩子:
执行后钩子:
hooks/pre-execute.cjshooks/post-execute.cjsContent Strategy Framework
内容策略框架
Step 1: Audience Mapping
步骤1:受众映射
Before writing a single word, define:
- Who: Primary audience segment (demographics, psychographics, intent signals)
- Where: Platform/channel they are on
- When: Buying stage / funnel position (Awareness → Consideration → Decision)
- Why: Pain point or aspiration driving search/scroll
Audience Awareness Stages → Framework Match:
| Awareness Stage | Best Framework | Why |
|---|---|---|
| Unaware (cold) | AIDA, PAS | Build attention first; introduce problem |
| Problem-aware | PAS, BAB | Lead with pain, show resolution |
| Solution-aware | 4Ps, FAB | Evidence-based; translate features to value |
| Product-aware | FAB, BAB | Show transformation; competitor contrast |
| Most aware | Direct CTA | Skip education; they are ready |
在动笔之前,先明确:
- Who:核心受众群体(人口统计特征、心理特征、意图信号)
- Where:他们活跃的平台/渠道
- When:购买阶段/漏斗位置(认知 → 考虑 → 决策)
- Why:驱动搜索/浏览的痛点或期望
受众认知阶段 → 框架匹配:
| 认知阶段 | 最优框架 | 原因 |
|---|---|---|
| 无认知(冷受众) | AIDA, PAS | 首先吸引注意力,引入问题 |
| 感知到问题 | PAS, BAB | 从痛点切入,展示解决方案 |
| 感知到解决方案 | 4Ps, FAB | 基于证据,将功能转化为价值 |
| 了解产品 | FAB, BAB | 展示转化效果,对比竞品 |
| 高认知 | 直接CTA | 跳过教育环节,用户已准备好转化 |
Step 2: Content Pillar Definition
步骤2:内容支柱定义
Organize content into 3-5 pillars per brand/product:
- Educational: Solves a problem (SEO-driven, evergreen)
- Inspirational: Shows aspiration, transformation (social-first)
- Promotional: Drives direct action (conversion-focused)
- Community: User-generated, testimonials (trust signals)
- Behind-the-Scenes: Builds brand personality (engagement)
每个品牌/产品的内容可分为3-5个支柱:
- 教育类:解决问题(SEO驱动,常青内容)
- 灵感类:展示期望、转变效果(社交优先)
- 推广类:驱动直接行动(聚焦转化)
- 社区类:用户生成内容、推荐语(信任信号)
- 幕后类:塑造品牌个性(提升engagement)
Step 3: Channel Selection
步骤3:渠道选择
| Channel | Content Type | Goal | Cadence |
|---|---|---|---|
| Blog/SEO | Long-form evergreen, 1500-3000 words | Organic traffic | 2-4x/month |
| Thought leadership, case studies | B2B awareness | 3-5x/week | |
| Segmented newsletters, drip sequences | Retention + conversion | 1-3x/week | |
| TikTok/Reels | Short-form video, <60s hooks | Discovery + top-of-funnel | Daily |
| Carousel, events, community posts | Community + retargeting | 3-5x/week |
| 渠道 | 内容类型 | 目标 | 发布频次 |
|---|---|---|---|
| 博客/SEO | 长文常青内容,1500-3000词 | 自然流量 | 每月2-4次 |
| 思想领导力内容、案例研究 | B2B认知 | 每周3-5次 | |
| 邮件 | 分群 newsletter、 drip 序列 | 留存+转化 | 每周1-3次 |
| TikTok/Reels | 短视频,<60秒钩子 | 发现+漏斗顶层 | 日更 |
| 轮播图、活动、社区帖子 | 社区+再营销 | 每周3-5次 |
Copywriting Patterns
文案写作模式
AIDA Framework (Cold Audience)
AIDA框架(冷受众)
ATTENTION: Grab with bold claim, stat, or question
INTEREST: Explain relevance to reader's situation
DESIRE: Show transformation, testimonials, proof
ACTION: Single clear CTA (avoid multiple options)Example (Email subject + body):
Subject: "68% of marketers are wasting their content budget"
[ATTENTION] Most companies publish content nobody reads.
[INTEREST] The difference? A documented content strategy.
[DESIRE] Teams using structured content plans see 3x ROI.
[ACTION] Download the 2025 Content Strategy Playbook →ATTENTION: 用大胆的观点、数据或问题吸引注意力
INTEREST: 说明和读者处境的相关性
DESIRE: 展示转变效果、推荐语、证明
ACTION: 单一明确的CTA(避免多个选项)示例(邮件主题+正文):
主题: "68%的营销人员正在浪费他们的内容预算"
[ATTENTION] 大多数公司发布的内容没人看。
[INTEREST] 区别在哪里?是否有书面的内容策略。
[DESIRE] 使用结构化内容规划的团队ROI可达3倍。
[ACTION] 下载2025内容策略手册 →PAS Framework (Problem-Aware Audience)
PAS框架(感知到问题的受众)
PROBLEM: Name the exact pain point
AGITATE: Amplify the consequences of inaction
SOLUTION: Present your offer as the logical answerPROBLEM: 点明具体痛点
AGITATE: 放大不行动的后果
SOLUTION: 把你的产品作为合理解决方案推出BAB Framework (Transformation Stories)
BAB框架(转变故事)
BEFORE: Describe life with the problem
AFTER: Paint the aspiration/desired state
BRIDGE: Explain how your product/service creates the bridgeBEFORE: 描述有问题的生活状态
AFTER: 描绘期望/理想状态
BRIDGE: 说明你的产品/服务如何搭建两者之间的桥梁4Ps Framework (Informed/Warm Audience)
4Ps框架(了解情况/暖受众)
PROBLEM: State the problem (brief; audience already aware)
PROMISE: Make a specific, believable claim
PROOF: Evidence (stats, case studies, testimonials)
PROPOSAL: Concrete offer with CTAPROBLEM: 说明问题(简短,受众已经知晓)
PROMISE: 给出具体可信的承诺
PROOF: 证据(数据、案例研究、推荐语)
PROPOSAL: 带CTA的具体报价FAB Framework (Product Copy)
FAB框架(产品文案)
FEATURE: What the product has/does
ADVANTAGE: Why that feature matters
BENEFIT: How it improves the customer's lifeFEATURE: 产品的功能/作用
ADVANTAGE: 该功能的价值
BENEFIT: 如何改善客户的生活Platform-Specific Content Guidelines
平台专属内容指南
TikTok / Instagram Reels
TikTok / Instagram Reels
- Hook in first 1-3 seconds — text overlay + strong visual
- Ideal length: 15-30s (watch-through rate drops sharply after 30s)
- Use trending audio (2x engagement lift vs. original audio)
- Captions: conversational, 3-5 sentences, include 3-5 relevant hashtags
- CTA: "Follow for more", "Comment your answer", "Link in bio"
- Content types: tutorials, trends, POV, behind-the-scenes
- 前1-3秒放钩子 — 文字叠加+强视觉
- 理想长度:15-30秒(30秒后完播率大幅下降)
- 使用热门音频(比原创音频engagement高2倍)
- 文案:口语化,3-5句话,包含3-5个相关标签
- CTA:"关注获取更多内容"、"评论留下你的答案"、"链接在简介"
- 内容类型:教程、热点、POV、幕后内容
- Long-form posts: 1200-1500 characters for maximum reach
- Hook (first line must standalone as preview): bold claim or question
- Structure: Hook → Story/Data → Insight → CTA
- Posting time: Tuesday–Thursday, 8-10am or 12-2pm (business timezone)
- Images: native documents (carousels) outperform external links by 3x
- Avoid: external links in post body (kills reach); put in comments
- 长文:1200-1500字符获得最大曝光
- 钩子(第一行必须可以独立作为预览):大胆的观点或问题
- 结构:钩子 → 故事/数据 → 洞察 → CTA
- 发布时间:周二-周四,8-10点或12-2点(工作时区)
- 图片:原生文档(轮播图)表现比外部链接好3倍
- 注意:不要在正文放外部链接(会降低曝光),放在评论区
邮件
- Subject line (A/B test mandatory): 40-60 characters, include power word
- Preview text: 85-100 characters, extends subject line promise
- Single CTA per email (multiple CTAs reduce click rate by 25%)
- Segmentation: separate sequences by funnel stage and behavior
- Cadence: max 3x/week; nurture sequences: 7-10 emails over 2-4 weeks
- Mobile-first layout: single column, 600px max width, 16px minimum font
- 主题行(必须做A/B测试):40-60字符,包含强吸引力词汇
- 预览文本:85-100字符,延伸主题行的承诺
- 每封邮件仅一个CTA(多个CTA会降低25%的点击率)
- 分群:按漏斗阶段和行为拆分序列
- 发布频次:每周最多3次;培育序列:2-4周发送7-10封邮件
- 移动端优先布局:单列,最大宽度600px,最小字号16px
Blog / SEO
博客 / SEO
- Primary keyword in H1, URL slug, first 100 words, at least 2 H2s
- Target featured snippet with direct answer paragraph (40-60 words)
- Internal links: minimum 3 relevant internal links per post
- Word count: 1500-3000 words for competitive SERP positions
- Schema markup: Article or HowTo depending on content type
- Update cadence: refresh top posts every 12-18 months
- 核心关键词放在H1、URL slug、前100词、至少2个H2中
- 用40-60词的直接回答段落争取精选片段
- 内部链接:每篇文章至少3个相关内部链接
- 字数:1500-300词可以获得有竞争力的SERP排名
- Schema标记:根据内容类型选择Article或HowTo
- 更新频次:每12-18个月更新top文章
Editorial Calendar Management
编辑日历管理
Calendar Structure
日历结构
Month View:
- Content pillars assigned to week blocks
- Platform rotation (ensures channel balance)
- Campaign anchors (product launches, seasonal events, holidays)
- Buffer capacity (20% reserved for reactive/trending content)
Week View:
- Monday: Brief writer + assign assets
- Tuesday-Wednesday: Draft creation
- Thursday: Review + edits
- Friday: Schedule/publish月视图:
- 内容支柱分配到周块
- 平台轮动(确保渠道平衡)
- 活动锚点(产品发布、季节活动、假日)
- 缓冲容量(20%留给热点/即时内容)
周视图:
- 周一:给作者下发brief + 分配素材
- 周二-周三:草稿创作
- 周四:审核+修改
- 周五:排期/发布Content Brief Template
内容Brief模板
markdown
undefinedmarkdown
undefinedContent Brief
内容Brief
Title/Working Headline: [H1 target]
Content Type: Blog / Email / Social / Ad
Platform: [channel]
Pillar: Educational / Inspirational / Promotional / Community
Framework: AIDA / PAS / BAB / 4Ps / FAB
Target Audience: [segment + awareness stage]
Primary Goal: [Awareness / Traffic / Lead / Conversion]
Primary Keyword/Topic: [keyword or topic]
CTA: [exact text + destination]
Due Date: YYYY-MM-DD
Assigned To: [human / AI / both]
Word Count / Length: [target]
Assets Needed: [images, video, graphics]
undefined标题/待定标题: [H1目标]
内容类型: 博客 / 邮件 / 社交 / 广告
平台: [渠道]
支柱: 教育类 / 灵感类 / 推广类 / 社区类
框架: AIDA / PAS / BAB / 4Ps / FAB
目标受众: [群体 + 认知阶段]
核心目标: [认知 / 流量 / 线索 / 转化]
核心关键词/主题: [关键词或主题]
CTA: [准确文本+跳转地址]
截止日期: YYYY-MM-DD
负责人: [人类 / AI / 两者]
字数/长度: [目标]
所需素材: [图片、视频、图形]
undefinedWorkflow States
工作流状态
- → Content brief drafted
IDEATION - → Draft being written
IN_PROGRESS - → Awaiting approval
REVIEW - → Approved + in queue
SCHEDULED - → Live
PUBLISHED - → Post-publish tracking window (7-30 days)
MEASURING
- → 内容brief已起草
IDEATION - → 草稿撰写中
IN_PROGRESS - → 等待审批
REVIEW - → 已审批+进入发布队列
SCHEDULED - → 已上线
PUBLISHED - → 发布后跟踪窗口(7-30天)
MEASURING
A/B Testing Workflow
A/B测试工作流
Test Design Protocol
测试设计协议
1. HYPOTHESIS: "Changing X to Y will increase Z because [reason]"
2. VARIABLE: Isolate ONE variable per test (subject line, CTA, headline, image)
3. SAMPLE: Minimum 500 impressions per variant for statistical significance
4. DURATION: Run minimum 7 days to account for day-of-week variance
5. METRIC: Define primary metric BEFORE running (CTR, conversion rate, open rate)1. 假设: "将X改成Y会提升Z,因为[原因]"
2. 变量: 每次测试仅隔离一个变量(主题行、CTA、标题、图片)
3. 样本: 每个变体至少500次曝光,保证统计显著性
4. 时长: 至少运行7天,消除星期几的差异影响
5. 指标: 运行前先定义核心指标(CTR、转化率、打开率)Test Backlog (Priority Order)
测试待办(优先级排序)
| Variable | Impact | Effort | Recommended Order |
|---|---|---|---|
| Email subject line | High | Low | 1st |
| Ad headline | High | Low | 2nd |
| CTA text | High | Low | 3rd |
| Landing page hero | High | Medium | 4th |
| Email send time | Medium | Low | 5th |
| Content format (video vs. static) | High | High | 6th |
| 变量 | 影响 | 工作量 | 推荐顺序 |
|---|---|---|---|
| 邮件主题行 | 高 | 低 | 1st |
| 广告标题 | 高 | 低 | 2nd |
| CTA文本 | 高 | 低 | 3rd |
| 落地页首屏 | 高 | 中 | 4th |
| 邮件发送时间 | 中 | 低 | 5th |
| 内容格式(视频vs静态) | 高 | 高 | 6th |
Test Result Logging
测试结果记录
json
{
"test_id": "email-subject-2026-03",
"variable": "subject_line",
"variant_a": "68% of marketers waste their budget",
"variant_b": "Is your content strategy costing you money?",
"metric": "open_rate",
"result_a": 0.24,
"result_b": 0.31,
"winner": "b",
"confidence": 0.95,
"applied_to": "all future campaign emails",
"date": "2026-03-01"
}json
{
"test_id": "email-subject-2026-03",
"variable": "subject_line",
"variant_a": "68% of marketers waste their budget",
"variant_b": "Is your content strategy costing you money?",
"metric": "open_rate",
"result_a": 0.24,
"result_b": 0.31,
"winner": "b",
"confidence": 0.95,
"applied_to": "all future campaign emails",
"date": "2026-03-01"
}Campaign Planning
活动策划
Campaign Structure
活动结构
CAMPAIGN NAME: [descriptive + date range]
OBJECTIVE: [SMART goal: Awareness / Traffic / Leads / Revenue]
AUDIENCE: [Primary segment + targeting parameters]
BUDGET: [total + channel allocation]
CHANNELS: [ranked by expected ROI]
TIMELINE: [start → warm-up → peak → wind-down → analysis]
KPIs: [primary metric + 2-3 supporting metrics]
CONTENT MAP: [content pieces by channel and funnel stage]活动名称: [描述性+日期范围]
目标: [SMART目标: 认知 / 流量 / 线索 / 收入]
受众: [核心群体+定位参数]
预算: [总预算+渠道分配]
渠道: [按预期ROI排序]
时间线: [开始 → 预热 → 峰值 → 收尾 → 分析]
KPIs: [核心指标 + 2-3个辅助指标]
内容地图: [按渠道和漏斗阶段划分的内容]Campaign Content Map (Example)
活动内容地图(示例)
| Funnel Stage | Channel | Content Type | Framework | Goal |
|---|---|---|---|---|
| Awareness | TikTok | 30s tutorial video | AIDA | Reach 50k |
| Awareness | Blog | SEO article | Educational | 1000 organic visits |
| Consideration | Drip sequence (5 emails) | PAS | 500 nurture enrollments | |
| Conversion | Offer email | 4Ps | 50 conversions | |
| Retention | Onboarding sequence | FAB | 80% activation rate |
| 漏斗阶段 | 渠道 | 内容类型 | 框架 | 目标 |
|---|---|---|---|---|
| 认知 | TikTok | 30秒教程视频 | AIDA | 触达5万人 |
| 认知 | 博客 | SEO文章 | 教育类 | 1000次自然访问 |
| 考虑 | 邮件 | Drip序列(5封邮件) | PAS | 500人加入培育序列 |
| 转化 | 邮件 | 优惠邮件 | 4Ps | 50次转化 |
| 留存 | 邮件 | 新手引导序列 | FAB | 80%激活率 |
Content Performance KPIs
内容表现KPIs
Tier 1: Engagement (Content Quality Signal)
层级1:互动(内容质量信号)
| KPI | Definition | Target Benchmark |
|---|---|---|
| Engaged sessions | Sessions >10s with interaction | >60% of sessions |
| Scroll depth | % of page scrolled | >50% to 75% mark |
| Time on page | Average seconds spent | Varies by content type |
| Social shares | Organic amplification | >1% of views |
| KPI | 定义 | 目标基准 |
|---|---|---|
| 互动会话 | 会话时长>10秒且有交互 | >60%的会话 |
| 滚动深度 | 页面滚动百分比 | >50%到75%的位置 |
| 页面停留时间 | 平均停留秒数 | 随内容类型变化 |
| 社交分享 | 自然扩散 | >1%的浏览量 |
Tier 2: Conversion (Business Impact Signal)
层级2:转化(业务影响信号)
| KPI | Definition | Target Benchmark |
|---|---|---|
| CTR (organic) | Click-through rate from SERP | >3% |
| Email open rate | Opens / delivered | >25% |
| Email CTR | Clicks / opened | >3% |
| Content conversion rate | CTA completions / visitors | >2% |
| KPI | 定义 | 目标基准 |
|---|---|---|
| CTR(自然) | SERP的点击率 | >3% |
| 邮件打开率 | 打开数/送达数 | >25% |
| 邮件CTR | 点击数/打开数 | >3% |
| 内容转化率 | CTA完成数/访客数 | >2% |
Tier 3: Retention & Loyalty
层级3:留存&忠诚度
| KPI | Definition | Target Benchmark |
|---|---|---|
| Return visit rate | % of visitors who return | >20% in 30 days |
| Email list growth rate | Net new subscribers / total | >5% monthly |
| Unsubscribe rate | Churned / sent | <0.5% |
| KPI | 定义 | 目标基准 |
|---|---|---|
| 回访率 | 回访访客占比 | 30天内>20% |
| 邮件列表增长率 | 净新增订阅者/总数 | 每月>5% |
| 退订率 | 流失数/发送数 | <0.5% |
Tier 4: ROI
层级4:ROI
| KPI | Definition | Target Benchmark |
|---|---|---|
| Content ROI | (Revenue attributable - cost) / cost | >200% |
| Cost per lead | Total content cost / leads | Varies by industry |
| Customer acquisition cost via content | Total cost / customers | Trending down |
| KPI | 定义 | 目标基准 |
|---|---|---|
| 内容ROI | (可归属收入-成本)/成本 | >200% |
| 每条线索成本 | 总内容成本/线索数 | 随行业变化 |
| 内容渠道获客成本 | 总成本/获客数 | 趋势下降 |
Iron Laws
铁律
- ALWAYS match copywriting framework to audience awareness stage — using AIDA for a product-aware audience wastes attention; using FAB for a cold audience loses them before they care.
- NEVER publish content without a measurable KPI defined first — content without a success metric cannot be optimized and cannot prove ROI.
- ALWAYS A/B test one variable at a time — testing multiple variables simultaneously makes it impossible to attribute performance changes to a single cause.
- NEVER use a one-size-fits-all content format across platforms — each platform's algorithm rewards native content formats; cross-posting without adaptation produces 50-80% lower organic reach.
- ALWAYS maintain human editorial oversight for AI-drafted content — AI handles research, structure, and drafts; human editors ensure brand voice, factual accuracy, and strategic alignment.
- 始终将文案框架和受众认知阶段匹配 — 对了解产品的受众用AIDA会浪费注意力;对冷受众用FAB会让他们在产生兴趣前就流失。
- 永远不要在未定义可衡量KPI的情况下发布内容 — 没有成功指标的内容无法优化,也无法证明ROI。
- 始终一次仅A/B测试一个变量 — 同时测试多个变量会导致无法将表现变化归因到单一原因。
- 永远不要跨平台使用通用内容格式 — 每个平台的算法都奖励原生内容格式;不做适配直接跨平台发布会降低50-80%的自然曝光。
- AI生成的内容始终要有人工编辑审核 — AI负责调研、结构和草稿;人工编辑保证品牌调性、事实准确性和战略对齐。
Anti-Patterns
反模式
| Anti-Pattern | Why It Fails | Correct Approach |
|---|---|---|
| Using AIDA for product-aware audience | Over-educates audience that already knows the problem | Use 4Ps or FAB; lead with proof and offer |
| Cross-posting identical content to all platforms | Platform algorithms penalize non-native formats | Adapt format, length, and tone per platform |
| No A/B testing before scaling ad spend | Intuition-based creative selection leaves 30-40% CTR on the table | Test headlines + CTAs first; scale winners |
| Measuring pageviews as content success | Pageviews measure traffic not content quality | Track engaged sessions and conversion rate |
| Publishing without a content brief | Inconsistent messaging, poor SEO, no clear CTA | Require brief for every piece before writing starts |
| Writing for search engines, not humans | High bounce rate, low engagement, penalized by Google | Write for humans first; optimize secondarily |
| 反模式 | 失败原因 | 正确做法 |
|---|---|---|
| 对了解产品的受众用AIDA | 对已经知道问题的受众做过度教育 | 用4Ps或FAB;从证据和优惠切入 |
| 所有平台发布完全相同的内容 | 平台算法会惩罚非原生格式 | 每个平台适配对应的格式、长度和语气 |
| 扩大广告支出前不做A/B测试 | 基于直觉的创意选择会浪费30-40%的CTR潜力 | 先测试标题+CTA;扩大优胜变体的投放 |
| 将页面浏览量作为内容成功指标 | 页面浏览量衡量的是流量不是内容质量 | 跟踪互动会话和转化率 |
| 没有内容brief就发布 | 信息不一致、SEO表现差、没有明确CTA | 每篇内容动笔前都要求有brief |
| 为搜索引擎写作而非为人类写作 | 高跳出率、低互动、被谷歌惩罚 | 首先为人类写作;其次做优化 |
Memory Protocol (MANDATORY)
记忆协议(强制要求)
Before starting:
Read
.claude/context/memory/learnings.mdAfter completing:
- New pattern →
.claude/context/memory/learnings.md - Issue found →
.claude/context/memory/issues.md - Decision made →
.claude/context/memory/decisions.md
ASSUME INTERRUPTION: If it's not in memory, it didn't happen.
开始前:
阅读
.claude/context/memory/learnings.md完成后:
- 新模式 →
.claude/context/memory/learnings.md - 发现问题 →
.claude/context/memory/issues.md - 做出决策 →
.claude/context/memory/decisions.md
假设会中断:如果没有记录在记忆中,就等于没发生过。