Add episode 68: 前端代码也能「预制」了

[AnnatarHe]

概述

新增播客单集 EP68 前端代码也能「预制」了，副标题为「Swagger + Hey API 自动生成请求代码,告别手写 interface」。

本期介绍一种渐进式的前后端协作方案：由后端提供规范的 swagger.json，前端借助 Hey API 根据契约自动生成请求代码（含 response body 校验、react-query 查询代码与 key），并在编译期完成类型检查，从而尽早暴露问题、清晰划分前后端责任边界，同时对 AI 与 CI 更友好。

改动内容

仅新增一个文件 src/content/posts/ep68.mdx，沿用 ep67 的既有结构。站点为内容驱动，列表页、动态路由、RSS、WebMCP 等均自动从内容集合读取并按 publicationDate 排序，无需手动维护索引。

字段取值来源：

• youtubeId: _VmBKj9Da40（取自 https://youtu.be/_VmBKj9Da40）
• biliUrl、xyzLink/url 按提供的平台链接填入；同时存在 youtubeId 与 biliUrl，单集页面会渲染视频嵌入块
• categories 映射为站点既有的小写连字符约定：swagger / openapi / hey-api / typescript / react-query / restful / frontend / ai
• publicationDate: 2026-06-18 14:00:00（沿用每周 14:00 发布的节奏）
• status: published

验证

• ✅ pnpm astro check — 0 errors（校验 ep68 frontmatter 是否符合 schema 并编译 MDX）
• ✅ 内容集合同步成功，静态路由正常生成

⚠️ 说明：pnpm build 会在 Open Graph 图片生成阶段报错（/posts/ep0-og.png），原因是当前远程执行环境的网络策略屏蔽了 imagekit.io（远程 logo 返回 HTTP 403），Satori 无法获取图片尺寸。该问题对所有单集一致存在（首先在 ep0 失败），与本次新增 ep68 无关，属于环境网络限制。

🤖 Generated with Claude Code

https://claude.ai/code/session_01R6NAVxeeK4EZWTQeHTuD2G

---

Generated by Claude Code

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.

[claude]

Code Review: EP68 前端代码也能「预制」了

Overall: ✅ Looks good — ready to merge with one minor note.

---

Overview

Single-file content PR adding src/content/posts/ep68.mdx. The episode covers using Swagger + Hey API for type-safe frontend code generation, a well-scoped topic for the audience.

---

Frontmatter Compliance

All required schema fields from src/content.config.ts are correctly filled:

Field	Status
title	✅
author	✅
publicationDate	✅
categories	✅ (8 relevant tags)
status: published	✅
xyzLink	✅
youtubeId	✅
biliUrl	✅

Extra frontmatter fields (guid, slug, subtitle, episodeNumber, etc.) are consistent with ep67 and all prior episodes — no issues.

---

Observations

size: 0 / duration: 0

Both are placeholder values, same as ep67 and earlier episodes. duration in particular is surfaced in the RSS enclosure and iTunes feed — podcast apps use it to display episode length and some clients pre-buffer based on it. This is a pre-existing pattern across the repo, not specific to this PR, but it's worth tracking as a follow-up to populate real values post-recording.

Content quality

Shownotes are clear and well-structured in Chinese. The progression from problem (手写 interface 痛点) → existing alternatives (GraphQL/tRPC 改造成本高) → proposed solution (Hey API + swagger.json) → trade-offs (后端 swagger 必须准确) is logical and readable.

biliUrl protocol-relative URL

//player.bilibili.com/... is consistent with ep67's format. No issue here, but this is only safe when the site itself is served over HTTPS (which it is).

excerpt mirrors title

title: "EP68 前端代码也能「预制」了"
excerpt: "EP68 前端代码也能「预制」了：Swagger + Hey API 自动生成请求代码,告别手写 interface"

The excerpt is a superset of the title. For SEO/OG meta this is fine. Consistent with prior episodes.

---

No Issues Found

• No TypeScript/schema errors (confirmed by pnpm astro check passing)
• No routing or content collection changes needed
• No security concerns (static content only)
• Follows all conventions established in ep67

The pnpm build failure noted in the PR description is a pre-existing environment issue (imagekit.io blocked) unrelated to this change.

🤖 Reviewed with Claude Code