[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"pack-detail-ai-side-hustle-kit-zh":3,"seo:pack:ai-side-hustle-kit:zh":99},{"code":4,"message":5,"data":6},200,"操作成功",{"pack":7},{"slug":8,"icon":9,"tone":10,"status":11,"status_label":12,"title":13,"description":14,"items":15,"install_cmd":98},"ai-side-hustle-kit","💼","#10B981","new","本周新建","AI 副业起步套装","十个资产，把你的 agent 配置成副业打工人 — 调研、爬数据、内容生产、外联、自动化胶水、记忆层全覆盖。选一套今晚就能跑，周日发出第一版。",[16,28,38,46,54,62,69,76,83,90],{"id":17,"uuid":18,"slug":19,"title":20,"description":21,"author_name":22,"view_count":23,"vote_count":24,"lang_type":25,"type":26,"type_label":27},25,"23330210-b26a-4d97-ad97-1735c203eaa6","gpt-researcher-autonomous-research-report-agent-23330210","GPT Researcher — Autonomous Research Report Agent","AI agent that generates detailed research reports from a single query. Searches multiple sources, synthesizes findings, and cites references.","TokRepo精选",572,0,"en","skill","Skill",{"id":29,"uuid":30,"slug":31,"title":32,"description":33,"author_name":34,"view_count":35,"vote_count":24,"lang_type":25,"type":36,"type_label":37},635,"f73611a0-142f-4364-97dc-b57eb03473ad","tavily-search-api-built-ai-agents-rag-f73611a0","Tavily — Search API Built for AI Agents & RAG","Search API designed specifically for AI agents and RAG pipelines. Returns clean, LLM-ready results with content extraction, no HTML parsing needed. Official MCP server available. 5,000+ stars.","Tavily",221,"mcp","MCP",{"id":39,"uuid":40,"slug":41,"title":42,"description":43,"author_name":44,"view_count":45,"vote_count":24,"lang_type":25,"type":26,"type_label":27},226,"a9cbbc61-0159-41a5-82a0-f44c24da8b55","jina-reader-convert-any-url-llm-ready-text-a9cbbc61","Jina Reader — Convert Any URL to LLM-Ready Text","Convert any URL to clean, LLM-friendly markdown with a simple prefix. Just prepend r.jina.ai\u002F to any URL. Handles JS-rendered pages, PDFs, and images. 10K+ stars.","Script Depot",6979,{"id":47,"uuid":48,"slug":49,"title":50,"description":51,"author_name":52,"view_count":53,"vote_count":24,"lang_type":25,"type":36,"type_label":37},3607,"6c45e855-cfad-5617-b5fb-3660436a827a","apify-mcp-server-8-000-web-scrapers-for-agents","Apify MCP Server — 8,000+ Web Scrapers for Agents","Apify MCP Server connects agents to Apify Actors via a hosted endpoint (mcp.apify.com) or local run, turning thousands of web scrapers into callable tools.","MCP Hub",104,{"id":55,"uuid":56,"slug":57,"title":58,"description":59,"author_name":60,"view_count":61,"vote_count":24,"lang_type":25,"type":26,"type_label":27},165,"b7ec9ae7-1144-42f0-9335-a5f40fbd6605","n8n-ai-native-workflow-automation-b7ec9ae7","n8n — AI-Native Workflow Automation","Open-source workflow automation with 400+ integrations and built-in AI capabilities. Build AI agents, RAG pipelines, and automation workflows with a visual editor.","n8n",278,{"id":63,"uuid":64,"slug":65,"title":66,"description":67,"author_name":60,"view_count":68,"vote_count":24,"lang_type":25,"type":36,"type_label":37},516,"4635d46c-cbf7-4be7-837e-95818241a46c","n8n-mcp-server-build-automations-ai-1-396-nodes-4635d46c","n8n MCP Server — Build Automations with AI, 1,396 Nodes","MCP server giving AI agents access to 1,396 n8n nodes and 2,709 workflow templates. Build and manage n8n automations through natural language.",182,{"id":70,"uuid":71,"slug":72,"title":73,"description":74,"author_name":52,"view_count":75,"vote_count":24,"lang_type":25,"type":36,"type_label":37},698,"cf74621a-0188-4222-8c9b-285873183b56","composio-250-tool-integrations-ai-agents-cf74621a","Composio — 250+ Tool Integrations for AI Agents","Connect AI agents to 250+ tools (GitHub, Slack, Gmail, Jira, etc.) with managed auth and natural language actions. Works with LangChain, CrewAI, and OpenAI.",84,{"id":77,"uuid":78,"slug":79,"title":80,"description":81,"author_name":22,"view_count":82,"vote_count":24,"lang_type":25,"type":26,"type_label":27},4297,"721d23c5-ffea-448c-b2a6-67c905855aad","claude-code-agent-content-marketer-721d23c5","Claude Code Agent: Content Marketer","Use this agent when you need to develop comprehensive content strategies, create SEO-optimized marketing content, or execute multi-channel content campaigns to drive engagement...",46,{"id":84,"uuid":85,"slug":86,"title":87,"description":88,"author_name":52,"view_count":89,"vote_count":24,"lang_type":25,"type":36,"type_label":37},3562,"1f9d6b56-3dd6-5ff8-83a2-4e2547a6f5e7","agenticmail-email-sms-infra-for-ai-agents-mcp","AgenticMail — Email + SMS Infra for AI Agents (MCP)","AgenticMail provides email + SMS infrastructure for AI agents, with a CLI setup wizard plus an MCP server so tools can send\u002Freceive messages under control.",50,{"id":91,"uuid":92,"slug":93,"title":94,"description":95,"author_name":96,"view_count":97,"vote_count":24,"lang_type":25,"type":26,"type_label":27},703,"96da1f40-1823-4d87-a84f-7d8269edeb24","mem0-memory-layer-ai-applications-96da1f40","Mem0 — Memory Layer for AI Applications","Add persistent, personalized memory to AI agents and assistants. Mem0 stores user preferences, past interactions, and learned context across sessions.","Mem0",720,"tokrepo install pack\u002Fai-side-hustle-kit",{"pageType":100,"pageKey":8,"locale":101,"title":102,"metaDescription":103,"h1":104,"tldr":105,"bodyMarkdown":106,"faq":107,"schema":123,"internalLinks":129,"citations":142,"wordCount":155,"generatedAt":156},"pack","zh","AI 副业起步套装 — 10 个资产把 agent 配成副业打工人","GPT Researcher \u002F Tavily \u002F Jina Reader \u002F Apify \u002F n8n \u002F Composio \u002F AgenticMail \u002F Mem0 — 一套 10 个资产，让上班族用 agent 把调研、爬数据、写内容、发外联自动跑起来。周六接线，周日发出第一版。","AI 副业起步套装 — 上班族的 Agent 流水线","十个资产，按真实副业流水线顺序排：先调研（搞清楚做啥），再找数据，再写内容，再外联，最后用 n8n 做胶水定时跑。周六接线，周日开始有人帮你打工。","## 这个 pack 包含什么\n\n这是给「白天上班、晚上想搞副业」的人准备的栈 — 不是推特上那种 50 个框架的购物清单。每个工具只做副业流水线里的一件事：搞清楚做啥、找 lead、写内容、发外联、然后趁你开 1-on-1 跑不掉的时候继续在云端跑。\n\n整套**有意做成 agent 驱动**。你周六花一下午接线，从周日开始 agent 帮你干苦力，只有真要做决策（定价、定位、需要真人回的邮件）时你才介入。**不需要先学三周 LangGraph**。选一个调研 agent、选一个自动化跑器、选一个外联通道 — 直接发车。\n\n## 推荐安装顺序\n\n1. **GPT Researcher** — 自主调研 agent。从这里起步，因为方向不对干啥都白搭。给它一句「\u003C我的想法>的竞品有哪些」或「谁真的在为 X 付费」，读完报告再敲第一行代码。\n2. **Tavily Search API** — 调研 agent 调用的搜索引擎。GPT Researcher 没有它也能跑，但 Tavily 为 LLM 调过的排序能让报告质量翻 2-3 倍，少烧很多 token 去看垃圾页。\n3. **Jina Reader** — `https:\u002F\u002Fr.jina.ai\u002F\u003Curl>` 直接返回任意页面的干净 markdown。这是最不起眼但你用得最频繁的工具：竞品页 \u002F Reddit 帖 \u002F 价格页直接喂 Claude\u002FGPT，省一个爬虫钱。\n4. **Apify MCP Server** — 8,000+ 现成爬虫（LinkedIn \u002F Twitter \u002F 大众点评 \u002F Google Maps 全有）以 MCP 工具方式暴露。Jina Reader 不够用、要**结构化** lead 列表时上这个。\n5. **n8n** — 工作流自动化运行时。这是你的 cron。触发器 + 分支 + 500 个集成 + 自托管一个月 $5 Hetzner 就够。下面所有东西最终都跑在 n8n 流里。\n6. **n8n MCP Server** — 把 n8n 的 1,396 个节点暴露给 coding agent。你说一句「每天从 Apify 抓数据然后给我邮件 lead」，它返回一个能跑的 JSON 工作流。省掉拖节点税。\n7. **Composio** — 250+ 预授权集成（Slack \u002F Notion \u002F Sheets \u002F Gmail \u002F HubSpot）给 agent 用。当 n8n 不合适（你想让 agent 自己决定动作，而不是流程决定）时上 Composio SDK。\n8. **Claude Code Agent: Content Marketer** — Claude Code 子 agent，把调研变成博客 \u002F 推特线程 \u002F 落地页文案。和 GPT Researcher 输出搭配，从头到尾产内容，不再像每个 AI 博客那样千篇一律。\n9. **AgenticMail** — agent 用的 email + SMS 基础设施（含 MCP server）。可以做冷外联（慎用）、交易确认邮件、入站解析 — agent 真的能**读到回信**并分流。\n10. **Mem0** — 记忆层。当你积累了 50 个 lead，agent 忘了哪些已经发过邮件时，你就需要 Mem0。跨会话保留上下文 = 把 demo 变成副业的分水岭。\n\n## 它们怎么协同\n\n```\n               ┌─ GPT Researcher ─┐\n               │   （做什么）       │\n               └──────┬───────────┘\n                      ▼\n         ┌─────── Tavily ──── Jina Reader ───┐\n         │   （搜索引擎）   （URL → markdown）  │\n         └──────────────┬─────────────────────┘\n                        ▼\n                    Apify MCP\n                （结构化 lead 列表）\n                        │\n                        ▼\n            ┌───────── n8n ─────────┐\n            │   （cron + 分支）        │\n            │   ↑                    │\n            │   └─ n8n MCP Server    │\n            │      （agent 帮你拼流）  │\n            └────┬──────────┬────────┘\n                 ▼          ▼\n         Content Marketer  AgenticMail\n           （写博客）        （发邮件）\n                 │          │\n                 └────┬─────┘\n                      ▼\n                    Mem0\n               （记住谁收过啥）\n             + Composio （其他一切动作）\n```\n\n关键的连接是 **n8n + Mem0**：n8n 排时间表，Mem0 保证 agent 不会因为忘了周一发过邮件就周二又给同一个人发一遍。其他 8 个工具都挂在这两侧。\n\n## 你会遇到的取舍\n\n- **n8n vs 自己写 Python cron** — n8n 搭得快、午休时在手机上也能改、过两个月你不记得脚本怎么写时还能看懂。自定义代码更灵活但它是副业项目的坟场。默认 n8n；只有 n8n 够不着的 API 才下沉到代码。\n- **Composio vs 直连集成** — Composio 帮你处理 250 个工具的 OAuth、重试、限流，agent 直接 `composio.execute('slack.send_message', ...)` 就行。代价：依赖他们家云。副业项目省时间值得。要做正经产品，再评估自托管 Mautic + 自己 OAuth 流。\n- **AgenticMail 冷邮件外联** — 技术上很简单，法律和声誉上风险很大。AgenticMail 从 day 1 就用在交易邮件 + **opt-in** 列表上。一旦你从自己在意的域名群发冷邮件，发件人声誉要重建好几个月。别。\n- **GPT Researcher vs 直接用秘塔\u002FPerplexity** — 单次问题秘塔更快。GPT Researcher 的优势在三个场景：(a) 要一份**带引用**能直接分享的报告 (b) 同样的调研要在 50 个公司\u002F主题上批量重跑 (c) 调研要喂给后续 agent 流水线。它是把调研系统化的工具，不是替代聊天。\n\n## 常见踩坑\n\n- **发车前花一周对比框架** — 这个 pack 的意义就是干掉这件事。选 10 个工具、周六接 3 个、周日发个东西。其他的可以等。\n- **没 Mem0 就把 agent 放着跑** — 你的 `send_email` 节点同一周会给同一个 lead 触发两次。Mem0（或者干脆一张 Postgres 表）在上 cron 之前是必备。\n- **从主域名发冷邮件** — 申请一个专门的发件子域名（`mail.yourthing.com`），warm up 两周再发任何可能被举报的内容。AgenticMail 让发送变简单，发件人声誉还是只能你自己背。\n- **把它当成「要学的栈」而不是「要用的栈」** — 你不需要搞懂 Jina Reader 的 reranker 怎么实现的。你只需要粘 URL 拿 markdown。忍住读每个依赖源码的冲动。\n- **wedge 没找到就开始搭 agent** — GPT Researcher 存在就是让你**先**做市场验证。如果报告说「没人为这个付钱」，再漂亮的 agent 管道也救不了。听报告的话。",[108,111,114,117,120],{"q":109,"a":110},"整套跑下来副业一个月真实成本多少？","现实基线：验证期 $0-30\u002F月。n8n 自托管一台 $5 Hetzner、Jina Reader 免费额度（100 万 token\u002F月）、Tavily 免费额度（1000 次搜索\u002F月）、Mem0 免费 hobby 版，覆盖任何副业的前 90 天。早期会显形的两个付费项：LLM 账单（Claude\u002FOpenAI，预算 $10-30\u002F月），以及 Apify credits（如果拉几千 lead 大概 $10\u002F月）。AgenticMail 和 Composio 低量都有慷慨的免费额度。",{"q":112,"a":113},"完全不写代码能跑起来吗？","基本可以。n8n 是无代码运行时，n8n MCP Server 让 coding agent 用自然语言**替你**搭流。你需要把 API key 粘到 n8n 的 credential 框、偶尔在节点里写一小段 JS 表达式，但不打开代码编辑器也能跑通「抓数据 → 发邮件」流水线。等你开始认真调 agent prompt，会一点 Python 或 TypeScript 就值回了。",{"q":115,"a":116},"为啥 n8n 和 Composio 都要 — 不是一回事吗？","n8n 是 workflow-first（触发器、分支、定时 — **流程**决定做什么）。Composio 是 agent-first（LLM 调用工具，**agent**决定做什么）。日常「抓这些站、写报告、邮件给我」就用 n8n。聊天式 agent 根据用户话来执行动作就用 Composio。正经的独立开发栈最后两个都会有：n8n 跑定时任务，Composio 进对话 agent 内部。",{"q":118,"a":119},"GPT Researcher 真比直接让 Claude 调研强吗？","单次问题，不强 — Claude 自带搜索更快。GPT Researcher 在三种情况下值：(a) 要一份**带引用**能拿出来分享的报告 (b) 同样的调研要在 50 个公司\u002F主题上批量重跑 (c) 调研要喂给自动化流水线下一步。它是把调研系统化的工具，不是替代聊天。",{"q":121,"a":122},"今晚就能开始的最小版本是啥？","三件套：Jina Reader（免费 + 不用注册，直接 `r.jina.ai\u002F\u003Curl>`）+ n8n（任何 VPS 上一条 Docker 命令）+ Mem0（免费 hobby 版）。就这三个能搭出「每天早晨爬竞品博客 + Claude 总结 + 写到一个 doc 里 + 记住哪些贴看过」。两小时跑通一个真 agent 管道。其他 7 个工具下周末再加，管道喊你加的时候再加。",{"@context":124,"@type":125,"name":13,"description":126,"numberOfItems":127,"inLanguage":128},"https:\u002F\u002Fschema.org","ItemList","十个 agent 资产，上班族一个周末就能接成副业流水线：调研、爬数据、内容、外联、自动化胶水。",10,"zh-CN",[130,134,138],{"url":131,"anchor":132,"reason":133},"\u002Fzh\u002Fmulti-agent","多智能体编排模式","副业流水线长大后，多 agent 编排是下一站",{"url":135,"anchor":136,"reason":137},"\u002Fzh\u002Fai-tools-for\u002Fautomation","AI Agent 自动化工具集","n8n 和 Composio 属于 TokRepo 上更大的自动化目录",{"url":139,"anchor":140,"reason":141},"\u002Fzh\u002Ffeatured","TokRepo 精选资产","这十个资产是更大的 agent-ready 精选目录的一部分",[143,147,151],{"claim":144,"source_name":145,"source_url":146},"GPT Researcher 能从一句 query 生成带引用的调研报告","GPT Researcher GitHub 仓库","https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher",{"claim":148,"source_name":149,"source_url":150},"n8n 是 fair-code 协议的工作流自动化工具，含 500+ 集成","n8n 官网","https:\u002F\u002Fn8n.io\u002F",{"claim":152,"source_name":153,"source_url":154},"Mem0 为 AI agent 和应用提供持久化记忆层","Mem0 官方文档","https:\u002F\u002Fdocs.mem0.ai\u002F",1980,"2026-05-22T12:00:00Z"]