AI 技能人工智能【免费下载链接】marketingskillsMarketing skills for Claude Code and AI agents. CRO, copywriting, SEO, analytics, and growth engineering.项目地址https://gitcode.com/GitHub_Trending/mar/marketingskills点击查看免费下载本篇技术指南以开源仓库 marketingskills 中directory-submissions技能的定位变体库Positioning Variations Library为核心系统讲解为什么不同目录Product Hunt、SaaS 目录、AI 目录、Agent/MCP 注册表、无代码目录、开发者社区、B2B 评测平台必须使用不同开头的产品描述以及如何为每个层级生成去重、高转化的提交文案。读完本文你将掌握一套可复制、可直接套用的七类目录提交模板、标签体系与避坑清单并理解其背后的搜索引擎与 AI 引擎查重机制。核心框架首句随目录层级而变目录受众对不同叙事框架framing的响应截然不同。永远不要把同一段产品描述复制粘贴到所有目录——AI 引擎会惩罚重复内容duplicate content而每种目录类型都会奖励不同的开头句。使用本变体库为每个层级生成专属变体时需要把以下占位符替换为真实值[product]、[category]、[competitors]、[use-case]、[audience]。各层级推荐的首句模式如下层级Tier首句模式Lead sentence pattern原因Why创业 / 发布Startup / launch[Product] is the easiest way to [outcome] for [audience].创始人快速扫描时最看重「结果清晰度」outcome clarity。SaaS 目录[Product] is the [differentiator] alternative to [competitors].捕获 [competitor] alternative 这类搜索意图。AI 目录[Product] uses [AI capability] to [outcome].TAAFT / Futurepedia 的受众明确想要 AI 工具。Agent / MCP[Product] is an MCP-native / agent-native [category].小众但意图极强直接排除竞争对手。无代码No-code[Product] lets you build [output] without code.受众看重速度而非技术深度。开发者工具[Product] is a [technical category] with [differentiator].开发者希望开头就有实质内容substance。B2B 评测平台[Product] helps [audience] [measurable business outcome].评测者想要 ROI 语言。这一框架在技能的 SKILL.md 中被归纳为「Rule 3: Positioning varies by directory type」其速查表从面向什么受众该以什么开头的角度再次印证了同一逻辑投放面开头主打原因创业目录结果Outcome受众是创始人关心产品做什么。SaaS 目录替代框架Alternative framing人们搜索 [competitor] alternative要在那里迎接他们。AI 目录AI 优先架构AI-first architectureTAAFT/Futurepedia 受众明确要 AI 工具。Agent/MCP 目录Agent/MCP 角度小众但意图极强是真正的护城河。无代码目录易用 强大Ease power受众看重构建速度胜过深度。开发者目录技术深度Technical depth开发者受众奖励技术实质。B2B 评测站ROI 使用场景买家要结果和案例研究。为什么重复内容会被惩罚从技能的评估用例evals.json 中 eval id2可以看到该技能的既定行为是当用户问能不能把同一段描述复制到所有目录时Agent 必须拒绝并解释 Rule 3明确说明Claude、ChatGPT、Perplexity 等 AI 引擎在交叉引用目录内容时会降低重复文案的排名权重。这也是整个变体库存在的根本原因目录不只是给搜索引擎爬虫看的也是 AI 引擎生成best [category] tools答案时的语料来源。模板一创业 / 发布目录Startup / Launch目标平台Product Hunt、BetaList、Fazier、Uneed、DevHunt、Microlaunch、OpenHunts、LaunchVault、Firsto、PitchWallTagline10 词以内The [differentiator] way to [outcome] for [audience].短描述60 字符[Outcome-focused one-liner with product name]长描述150 词[Product] is the easiest way to [outcome] for [audience]. Built for teams who [pain point], [product] removes [friction] by [how].Unlike [competitor category], [product] [key differentiator 1] and [key differentiator 2]. You can [action 1] in under [timeframe], [action 2] without [limitation], and [action 3] that would normally require [cost or technical skill].We built [product] because [founder origin story in one sentence]. Its now used by [audience examples] to [use case examples].Try it free at [url]. No credit card, no setup.标签[product category], [audience type], [use case 1], [use case 2], [differentiator], [tech]实战要点创业目录的读者是其他创始人他们不关心技术架构只想知道这东西到底能帮我达成什么。模板特意安排了创始人故事段落founder origin story——在 SKILL.md 的 Prepare asset variations 步骤中创始人故事被定义为 2–3 句话的标准资产和 tagline、60 字符短描述、150 词长描述、5–8 个分类标签并列说明它是发布类目录的必备素材。模板二SaaS / 软件目录目标平台AlternativeTo、SaaSHub、G2、Capterra、GetApp、SourceForge、Slashdot、Startup Stash、F6STaglineThe [differentiator] alternative to [top competitors].长描述[Product] is a [differentiator] alternative to [competitor 1], [competitor 2], and [competitor 3] — built for [audience] who need [gap the competitors dont fill].Where [competitor 1] [limitation 1] and [competitor 2] [limitation 2], [product] [solves]. You get [feature 1], [feature 2], and [feature 3] in a single workspace, at [pricing relative to competitors].Key features: • [Feature 1] — [benefit] • [Feature 2] — [benefit] • [Feature 3] — [benefit] • [Feature 4] — [benefit] • [Integration 1], [Integration 2], [Integration 3] integrationsTrusted by [audience examples]. Start free at [url].标签[competitor] alternative, [category], [audience], [differentiator], [top 3 features]实战要点这一层级的核心是替代品叙事。SaaSHub 本身就是为 [tool] alternatives 类查询而排名的见 directory-list.md Tier 2 中的说明AlternativeTo 的正确提交方式就是作为你 Top 4 竞争对手的替代品提交。注意一个关键纪律不要编造竞争对手的缺陷——AI 引擎会交叉核验竞争功能声明并降低说谎页面的排名该规则同时出现在 SKILL.md 的 Rule 2 与本文档的 DONT 列表中。模板三AI 目录目标平台TAAFT、Futurepedia、Toolify、Future Tools、aitools.inc、AIStage、LogicBalls、SaasAIToolsTaglineAI-powered [category] for [audience].长描述[Product] is an AI-powered [category] that [core AI capability]. It uses [specific models / techniques] to [outcome] — so [audience] can [job to be done] in a fraction of the time.What makes it AI-first: • [AI feature 1] — [what it does] using [model/approach] • [AI feature 2] — [what it does] • [AI feature 3] — [what it does] • [AI feature 4] — [what it does][Product] is built on [tech stack] and supports [models/providers]. Use cases: [use case 1], [use case 2], [use case 3], [use case 4].Free tier available. No API keys required to start.标签AI [category], [AI capability 1], [AI capability 2], AI for [audience], [use case 1], [use case 2], [LLM provider], [differentiator]实战要点与 SaaS 目录不同AI 目录的受众TAAFT 月流量 2MFuturepedia 收录 5,000 工具是带着我要找一个 AI 工具的明确意图来浏览的所以必须以 AI 优先AI-first开头而不是先讲业务价值。模板要求具体写出所用模型/技术路线specific models / techniques这是与通用 SaaS 描述最大的差异点。该层级在 directory-list.md 中对应 Tier 3AI Tool Directories只在产品确实是 AI 原生时适用。模板四Agent / MCP 注册表目标平台Glama、APITracker、Linux Foundation MCP Registry、AI Agents List、AI Agent Store、AgentHunterTaglineMCP-native [category] for AI agents.长描述[Product] is an MCP-native [category] that lets AI agents [capability]. It exposes [MCP server capabilities] via the Model Context Protocol, so agents in Claude, ChatGPT, Cursor, and any MCP-compatible client can [actions].MCP capabilities: • [Tool 1] — [what the agent can do] • [Tool 2] — [what the agent can do] • [Tool 3] — [what the agent can do] • [Resource 1] — [context surfaced] • [Prompt 1] — [pre-built prompt]Authentication: [auth method]. Transports: stdio, HTTP, SSE. Security: [security posture].Installation: [one-line install command]. Docs: [docs URL].标签MCP, MCP server, AI agent, agent [category], Claude integration, Model Context Protocol, [domain], [auth type]实战要点这是最技术化、最挑剔的模板MCP 能力清单必须按 Tool / Resource / Prompt 三类分别列出这是 Model Context Protocol 规范下的标准能力模型同时要交代认证方式、传输方式stdio、HTTP、SSE、安全姿态和一行安装命令。这类注册表Glama 收录 20K 经过安全评级的 MCP 服务器Linux Foundation MCP Registry 采用 PR 制提交是传统 SaaS 产品无法进入的领域——正如 directory-list.md Tier 4 所注如果产品暴露 agent 能力或 MCP 服务器这是一块真正的护城河。模板五无代码目录目标平台NoCodeFinder、No Code MBA Tools Directory、We Are No Code、NoCode.TechTaglineBuild [output] without code.长描述[Product] lets you build [output] without writing code. Drag, drop, or describe what you want and [product] handles the rest — [technical concept 1] and [technical concept 2] are automatic.What you can build: • [Example project 1] — built in [timeframe] • [Example project 2] — built in [timeframe] • [Example project 3] — built in [timeframe]No-code friendly features: • [Visual feature 1] • [Visual feature 2] • [AI-assisted feature] • [Pre-built templates]Start free. No credit card. Templates included.标签no code, no-code [category], visual [tool], drag and drop, [output type], [audience type]实战要点无代码受众的核心诉求是快速构建出结果所以模板用可以构建什么 多久建成Example project — built in [timeframe]来呈现价值并把技术概念[technical concept 1] 和 [technical concept 2] are automatic描述为自动完成的黑盒。这类描述刻意弱化技术深度与开发者模板形成鲜明对比。模板六开发者 / 技术目录目标平台DevHunt、Stackshare、GitHub、Dev.to、Hacker News Show HNTagline[Technical category] with [technical differentiator].长描述[Product] is a [technical category] built on [tech stack]. It solves [technical problem] by [technical approach].Architecture: • [Component 1] — [tech used] • [Component 2] — [tech used] • [Component 3] — [tech used]Why its different: [technical insight or novel approach]. We chose [trade-off] because [reason].Open source: [yes/no/partial]. Self-hostable: [yes/no]. License: [license].API: [REST / GraphQL / MCP / gRPC]. SDKs: [languages]. Docs: [url].标签[language], [framework], [category], open source, API, [tech stack component], [architecture approach]实战要点这是唯一要求交代架构组件与技术选型权衡trade-off的模板。开发者目录尤其是 Hacker News Show HN对营销腔零容忍——directory-list.md 明确指出 Show HN 对夸大宣传有版主死亡惩罚moderator death penalty for hype且只在有真正技术角度架构、DSL、新颖方法时才提交。模板中的 Open source / Self-hostable / License / API / SDKs 字段正是 Hacker News 与 Stackshare 社区判定一个项目是否真实的标尺。模板七B2B 评测平台目标平台G2、Capterra、TrustRadius、GetApp、Gartner Digital Markets、CrozdeskTagline[Business outcome] for [audience].长描述[Product] helps [audience] [achieve measurable business outcome]. Teams use it to [use case 1], [use case 2], and [use case 3] — reducing [metric] by [percentage] and increasing [metric] by [percentage].Key benefits: • [Business benefit 1] with [how measured] • [Business benefit 2] with [how measured] • [Business benefit 3] with [how measured]Integrations: [enterprise integrations — HubSpot, Salesforce, Slack, etc.]Security: [SOC 2 / GDPR / compliance posture]. Support: [support tier]. Pricing: [pricing range].Trusted by [customer logos / company size]. Case studies at [url].标签[business use case], [vertical], [audience role], [compliance], enterprise [category], [integration 1]实战要点B2B 评测平台G2 与 Capterra 是采购决策场景的读者是采购方而非创作者模板因此强制要求可测量的商业结果reducing [metric] by [percentage]并额外要求安全合规姿态SOC 2 / GDPR、企业级集成HubSpot、Salesforce、Slack与定价区间。注意此模板中的数字必须是真实数据——SKILL.md 的 Reviews Playbook 提到 G2/Capterra 的 listing没有评论就毫无价值评测平台的权威性建立在可验证事实上。分类标签库Category Tag Library每次提交从相关板块中抽取5–8 个标签。绝不在同一层级下的两个目录中重复使用完全相同的标签组合——标签是内容指纹的一部分重复会进一步强化查重信号。通用Universal[category], [audience], [differentiator], [use case], AI, no-code, SaaS, [tech stack]行业IndustryB2B, B2C, DTC, ecommerce, fintech, edtech, healthtech, martech, devtools, productivity, creator tools, agency tools待办工作Job-to-be-donelead generation, lead qualification, customer onboarding, product recommendation, sales enablement, marketing automation, survey, assessment, calculator, quiz, intake formAI 专用AI-specificAI agent, LLM, generative AI, conversational AI, RAG, MCP, agent framework, AI form, AI quiz, AI assistant, AI automation技术类Technicalopen source, self-hosted, API-first, webhook, Zapier, no-code, low-code, embeddable, white-label, multi-tenant, SSO, SAML为什么要精心设计标签模板在 DONT 列表中的最后一条点明了原因——不要跳过标签列表这是版主把你路由到正确分类的方式。标签不仅影响 SEO还直接决定目录版主的分类审核结果用错标签会导致 listing 被拒或归入错误分类浪费首次提交优势triage rule见 SKILL.md。Do / Dont 快速参考应该做DO跨层级变化开头句Vary the opening sentence across tiers——这是变体库的核心操作。使用真实数字与具体差异点Use real numbers and specific differentiators。让语气匹配受众Match tone to audience技术语气给开发者、商务语气给 G2、兴奋语气给 Product Hunt。在创业目录中包含创始人/起源角度Include a founder/origin angle in startup directories。在 AI 目录中以 AI 优先角度开头Lead with the AI-first angle in AI directories。不应该做DONT不要到处复制粘贴同一段 150 词描述Copy-paste the same 150-word description everywhere。不要用模糊主张Vague claims如 blazing fast、game-changing。不要罗列所有功能——每个层级挑 3–5 个并轮换使用。不要编造竞争对手功能Lie about competitor features——AI 引擎会交叉核验并降权。不要跳过标签列表——它是版主路由分类的依据。在实战流程中应用变体库变体库不是孤立文档它在 directory-submissions 技能的完整工作流中承担文案弹药库的角色。将其放入执行流程后实际的操作闭环是这样的1. 按层级选择模板directory-list.md 将目录划分为 13 个层级Tier 1 旗舰发布平台、Tier 2 创业/SaaS 目录、Tier 3 AI 目录、Tier 4 Agent/MCP 注册表、Tier 5 无代码目录、Tier 6 Best of 榜单外联、Tier 7 集成市场、Tier 8 个人资料与内容平台、Tier 9 本地商业目录、Tier 10 论坛与社区、Tier 11 新闻稿与文章站、Tier 12 社交书签、Tier 13 垂直行业目录每一层级的受众与意图不同正好对应变体库中的模板Tier 1 / Tier 2 的发布与创业类目录 → 模板一Startup / LaunchTier 2 的 SaaS/软件目录 → 模板二SaaS / SoftwareTier 3 → 模板三AI DirectoriesTier 4 → 模板四Agent / MCP RegistriesTier 5 → 模板五No-CodeTier 1/2 中的开发者向平台DevHunt、Stackshare、Show HN→ 模板六Dev / TechnicalTier 2 中的 G2、Capterra、TrustRadius、GetApp、Crozdesk → 模板七B2B Review Platforms2. 用跟踪表记录每个变体submission-tracker-template.csv 为每次提交预置了Positioning Variant Used使用的定位变体与Tags Used使用的标签两列以及Submission Date、Status、Live URL、Backlink Verified、Account Email、Notes等字段。这意味着每个目录用哪个变体、哪组标签是可审计、可复用的——当同层级下两个目录需要差异化时查表即可知道已经用过哪些开头句与标签组合避免无意识重复。表内还预填了每个目录所属的 Tier 与 CategoryStartup / SaaS / AI / Agent / MCP / No-Code / Dev / B2B review / Local / Forum / Article / PR / Profile 等与变体库的七类模板一一对应。3. 按批次提交并核验SKILL.md 的 Batch Submit 步骤给出了每次提交的六步操作复制该层级对应的定位变体 → 填表 → 上传素材Logo、截图、演示视频→ 提交 → 记录日期/URL/状态/版主备注 →listing 上线后用curl -sIL https://directory.com/your-listing | grep -i rel核验外链是否为 dofollow。如果该命令的返回值中不含rel即没有relnofollow或relugc则该链接是 dofollow 外链。directory-list.md 的 Verification 章节补充了两种核验途径手动 Inspect 检查rel属性、以及用 Ahrefs Site Explorer 按目录域名过滤 Backlinks并建议每季度复核一次部分目录可能无预警地把所有出链改为 nofollow。4. 被评估用例验证的行为evals.json 中 eval id2 明确要求当用户提出把同一段描述复制到所有目录时Agent 应拒绝并解释 Rule 3、说明 AI 引擎会惩罚重复内容、列出不同投放面的不同框架、指定各变体长度10 词内 tagline、60 字符短描述、150 词长描述、5–8 个标签并引用 positioning-variations.md。这从技能自检层面锁定了变体库的使用纪律——它不是一份可选的文案模板而是该技能输出的强制组成部分。关联技能与生态位置定位变体库服务于更完整的增长链路在 README.md 的技能分类中directory-submissions属于 Paid Distribution付费与分发类别。目录外链的价值最终落在以下配套技能构建的落地页上SKILL.md 的 Destination Pages Strategy 强调先建落地页再提交目录让链接权值有处可落competitors—/alternatives/[tool]竞品替代页模式是目录替代品叙事的承接页也是转化率最高的目录落地页5–15%。programmatic-seo— 规模化生成替代页、集成页、模板页让每条 dofollow 外链都指向可转化的程序化页面。ai-seo— GEOGenerative Engine Optimization单 H1、顺序标题层级、FAQ Schema、对比表格让目录和落地页更易被 AI 引擎引用。schema—Organization、Product、SoftwareApplication、FAQPageJSON-LD 结构化数据是 GEO 引用率的技术底座。把这些串起来看目录提交是用每种目录一种定位变体的方式铺设外链与发现面而变体库保证这条铺设路径上的每一块砖都各不相同、精准命中各自受众——这正是它在本仓库分发体系中的价值所在。赞分享AI 技能人工智能【免费下载链接】marketingskillsMarketing skills for Claude Code and AI agents. CRO, copywriting, SEO, analytics, and growth engineering.项目地址https://gitcode.com/GitHub_Trending/mar/marketingskills点击查看免费下载相关推荐Spaceship 按目录配置指南用 direnv 为每个目录定制 Zsh 提示符Spaceship 按目录配置指南用 direnv 为每个目录定制 Zsh 提示符 Spaceship Prompt 的全部选项都是环境变量这为按目录配置开发工具Spaceship Prompt 按目录配置指南借助 direnv 为指定目录定制提示符行为Spaceship Prompt 按目录配置指南借助 direnv 为指定目录定制提示符行为 direnv 是一款围绕 .envrc 文件工作的环境变量管理工开发工具为 Marketing Skills 贡献 Agent Skill从目录创建到合并的完整实操指南为 Marketing Skills 贡献 Agent Skill从目录创建到合并的完整实操指南 Marketing Skills 是一个面向 ClaudeAI 技能人工智能上一篇dio插件生态cookie_manager到web_adapter全解析下一篇容器化应用依赖隔离案例gh_mirrors/do/dockerfiles pipenv实战创作声明：本文部分内容由AI辅助生成（AIGC），仅供参考