What checks does npm run prepare-release actually run?

13 serial checks: clean git status, required package.json fields, dependency integrity, npm audit clean of critical/high, version not yet published, package.json and lockfile versions match, CHANGELOG entry for the current version, TypeScript compile + tests, .d.ts emitted, native-code tests (if any), build artifact validation, package size warning over 2MB, and an MCP server smoke test (one JSON-RPC handshake).

What should I look for in an npm publish --dry-run?

Inspect the tarball file list: it must include dist/, package.json, README.md, CHANGELOG.md, LICENSE, and binaries (if any); it must never contain src/, tests/, tsconfig.json, or .env. If sources slip in your .npmignore or the files field in package.json is misconfigured — shipping source bloats the package and leaks internals, and once published you cannot recall it.

Should I ship MCP tools as beta first or go straight to stable?

Beta first: npm publish --tag beta and let early users test with npx [pkg]@beta install. The biggest risk for an MCP tool is weird behavior across OS and Node versions, and beta is your window to surface those in real environments — once stdio stays clean, the info command works, and native deps install cleanly, then npm publish promotes to stable.

What does the prepublishOnly script actually do?

Setting "prepublishOnly": "npm run build" in package.json makes npm run the build automatically before every publish, eliminating the classic mistake of shipping without rebuilding. Without it, MCP packages routinely go out with last version's dist/, and users hit import errors on first run; with it, even a flaky CI cannot push out a half-built package from your local.

What goes wrong if I forget to tag the release in git?

A tag is the anchor that binds a git commit to an npm version; without it nobody can confidently say which commit v1.2.3 came from, and rollback / audit chains break. The flow is: after npm publish succeeds, run git tag vX.Y.Z and git push origin vX.Y.Z, then drop the matching CHANGELOG section into a GitHub Release — npm version, git history, and release notes stay in lockstep.

MCP Publishing Guide - AI Engineer

学AI来匠人

📖

AI Engineer

50 chapters

·Introduction

01 基础与模型5

01
LLM API 入门
30分钟
02
AI 模型对比参考
20分钟
03
机器学习 / 深度学习基础速通
30分钟
04
Transformer 与注意力机制入门
30分钟
05
Prompt Engineering
45分钟

02 上下文与 RAG6

06
Context Fundamentals
30分钟
07
RAG 系统入门
60分钟
08
🔒LangChain 框架
60分钟
09
Context Engineering & Memory
35分钟
10
Context Compression & Optimization
35分钟
11
Context Degradation Patterns
30分钟

03 工具与能力5

12
Function Calling & Tool Use
35分钟
13
Tool Design Principles
35分钟
14
多模态与工具链
40分钟
47
Claude API 进阶 Tool Use：三招降本提准
40分钟
48
MCP 的 Code Execution 模式：把工具当代码库调用
25分钟

04 智能体构建8

15
🔒AI Agent 开发基础
90分钟
16
System Prompt 设计实战
45分钟
17
System Prompt 案例库
30分钟
18
Agent 框架速览
40分钟
19
Agent 框架横向对比
45分钟
20
Multi-Agent Patterns
35分钟
21
Agent 深入
45分钟
49
长任务 Agent 的 Harness 设计模式
25分钟

05 评测与工程化5

22
LLM 评测与质量监控：可运行的回归检查
45分钟
23
LLM-as-a-Judge 评测
35分钟
24
Synthetic Data & Augmentation
30分钟
25
LLM 工程化与网关
35分钟
26
工作流与自动化
40分钟

06 安全与部署6

27
安全与威胁建模
35分钟
28
Data Governance & Privacy
30分钟
29
部署与成本优化
35分钟
30
🔒生产环境部署
90分钟
31
Debugging & Incident Playbook
35分钟
50
Claude Code 沙箱安全机制：双层隔离
25分钟

07 AI Coding & 进阶实战13

32
AI Coding 工作流
40分钟
33
Vibe Coding 实战
45分钟
34
AI Rules 配置实战
30分钟
35
🔒MCP 开发最佳实践
45分钟
36
🔒MCP 发布指南
25分钟
37
Skills 范式：从 Agent 到技能库
35分钟
38
Claude Code 技术原理：上下文与记忆管理
45分钟
39
Claude Code Examples
30分钟
40
AI 产品与体验
35分钟
41
🔒AI 规则持续改进
30分钟
42
PR Review 方法论
20分钟
43
代码分析选项
15分钟
44
GitHub Issue 分析
15分钟

08 模型层：微调与开源部署2

45
LoRA / QLoRA / PEFT 微调原理与取舍
35分钟
46
开源模型、本地部署与模型路由
30分钟

Resources

🗺️ Learning Roadmap
📚 Wiki Docs
📋 Cheat Sheets

Home/Learn/AI Engineer/MCP Publishing Guide

🎯Learn More Efficiently

Keep your learning uninterrupted

✏️
Study Notes
Take notes while learning, synced to cloud
📈
Progress Tracking
Auto-save, resume anytime
🏅
Completion Certificate
Shareable learning achievement
💬
Community Q&A
Connect with 2000+ learners

Sign Up Free→Already have an account? Sign in

👥2,847 learners active

AI学习面试课程学习中心