Next Public Content Expansion Plan

Generated at: 2026-06-12

This plan is for human review and publishing coordination. It does not mark drafts for review, publish articles, or claim traffic.

Current Evidence

• Public articles: 15
• Live missing from sitemap: 0
• Searchability score: 100
• Search Console / analytics traffic evidence: not available in repo reports
• Broad AI demand clusters without public coverage: 8
• Review-ready drafts available: 633
• AI deployment review pack: 10 safe candidates
• Industry prompt review pack: 12 safe candidates
• Popular prompt bridge: 27 candidates, 0 publish-confirm commands

Expansion Goal

Move the site from a small Codex/Vercel-heavy public surface to a broader AI tools guide:

• AI deployment: large model deployment, API routing, keys, rate limits, smoke checks
• Agent deployment: tool calling, MCP, permissions, logs, human approval
• Memory and RAG: knowledge bases, vector search, citations, privacy boundaries
• Office AI: PPT, Excel/table cleanup, weekly reports, meeting notes
• Industry prompts: sales, customer service, operations, ecommerce, HR, education, finance

First Review Batch

These should be checked first because they cover broad search intent and currently have no matching public article.

| Priority | Area | Candidate | File | | ---: | --- | --- | --- | | 1 | Prompt library | 全行业 AI 提示词模板库怎么做：销售、运营、客服、HR、财务和教育都能用 | content/blog/industry-ai-prompts-template-library-2026.mdx | | 2 | Agent deployment | AI Agent 部署怎么做：用 Vercel AI SDK 理解工具调用、多步执行和上线检查 | content/blog/ai-agent-deployment-vercel-ai-sdk-guide.mdx | | 3 | Memory / RAG | AI Agent 记忆和 RAG 怎么设计：短期记忆、长期记忆、引用来源和隐私边界 | content/blog/ai-agent-memory-rag-design-guide.mdx | | 4 | Customer service | 客服 AI 提示词模板：回复草稿、工单分类、情绪安抚和升级判断 | content/blog/customer-service-ai-prompts-guide.mdx | | 5 | Sales | 销售 AI 提示词模板：客户画像、跟进话术、异议处理和会议纪要 | content/blog/sales-ai-prompts-guide.mdx | | 6 | Ecommerce | 电商 AI 提示词模板：商品标题、详情页、评价分析和售后回复 | content/blog/ecommerce-ai-prompts-guide.mdx | | 7 | Data / Excel | 数据分析 AI 提示词模板：指标解释、SQL 思路、异常排查和报告摘要 | content/blog/data-analysis-ai-prompts-guide.mdx | | 8 | LLM deployment | 大模型部署怎么选：Hugging Face Inference Endpoints、API、私有化和成本检查 | content/blog/llm-deployment-huggingface-inference-endpoints-guide.mdx | | 9 | API security | AI API Key 怎么安全管理：环境变量、权限、轮换、泄露应急 | content/blog/ai-api-key-security-rotation-guide.mdx | | 10 | No-code AI | Dify Workflow 和 Agent 怎么选：固定流程、工具调用和人工审核 | content/blog/dify-workflow-vs-agent-guide.mdx | | 11 | MCP / security | MCP Server 怎么部署才安全：本地、远程、权限、日志和工具白名单 | content/blog/mcp-server-deployment-security-checklist.mdx | | 12 | Office AI | AI 做 PPT 怎么开始：新手从大纲到成稿的流程 | content/blog/ai-ppt-beginner-guide.mdx |

Second Review Batch

Use these after the first batch has been checked for factual freshness and internal-link conflicts.

| Area | Candidate | File | | --- | --- | --- | | Operations prompts | 运营 AI 提示词模板：周报、SOP、活动计划、复盘和数据解释 | content/blog/operations-ai-prompts-guide.mdx | | HR prompts | 人力招聘 AI 提示词模板：JD、简历初筛、面试题和培训材料 | content/blog/hr-recruiting-ai-prompts-guide.mdx | | Education prompts | 教育 AI 提示词模板：备课、教案、测验、反馈和学习计划 | content/blog/education-ai-prompts-guide.mdx | | Finance prompts | 财务 AI 提示词模板：报表摘要、费用分类、预算复盘和风险清单 | content/blog/finance-ai-prompts-guide.mdx | | Product prompts | 产品经理 AI 提示词模板：需求分析、PRD、竞品、用户故事和验收标准 | content/blog/product-manager-ai-prompts-guide.mdx | | Marketing prompts | 营销 AI 提示词模板：选题、广告文案、SEO 和活动复盘怎么写 | content/blog/marketing-ai-prompts-guide.mdx | | n8n RAG memory | n8n AI Agent 怎么接知识库和记忆：RAG、上下文和状态存储 | content/blog/n8n-ai-agent-rag-memory-guide.mdx | | Dify errors | Dify 工作流怎么做错误处理：变量、分支、重试和人工兜底 | content/blog/dify-workflow-error-handling-guide.mdx | | Model serving | 大模型部署成本和延迟怎么估算：上线前检查清单 | content/blog/llm-serving-cost-latency-checklist.mdx | | Table cleanup | AI 表格整理、清洗、分类和复盘相关内容应和 /tools/spreadsheet-cleaner 互链 | existing draft pool |

Review Rules

• Keep every candidate as draft/noindex until a human approves it.
• Do not run mark:review without explicit approval.
• Do not run publish confirm without separate explicit approval.
• Verify current official docs for APIs, SDK names, model names, pricing, rate limits, deployment commands, and platform policies.
• Remove or rewrite any claim about traffic, income, rankings, conversion, or guaranteed results.
• For Agent/MCP/RAG articles, require permissions, human approval, logs, privacy, citations, and rollback boundaries.
• For prompt articles, require input fields, output format, quality checks, negative examples, and human escalation rules.

Practical Next Step

If the first batch passes human review, it can raise the public article count from 15 toward 25-27 without changing the site strategy into a pure service site. The positioning remains: useful AI tools, tutorials, templates, and reviewable workflows first; monetization tools later.