# Bruce on AI Engineering — Full Content Index
# https://www.heyuan110.com/
# Generated: 2026-09-17
# 436 articles, bilingual (English default, Chinese under /zh/)


# ===== English Articles =====

## AI — 179


- [Modern Windows Shell Guide 2026: PowerShell 7 & Nushell](https://www.heyuan110.com/posts/ai/2026-09-17-modern-windows-shell-terminal-guide/): Stop mistaking terminals for shells. In 2026, the ultimate Windows CLI setup pairs PowerShell 7 for everyday development, Nushell for structured data, and WSL2.
- [Configuring Third-Party Models in OpenAI Codex: The Ultimate TOML & CC Switch Guide](https://www.heyuan110.com/posts/ai/2026-09-15-configuring-third-party-models-in-codex/): Unlock third-party models like DeepSeek & Claude in OpenAI Codex. Complete guide covering custom config.toml providers, separate profiles, and CC Switch protocol routing.
- [The Triumph of Minimalism: A Deep Dive into Pi (pi.dev) Agent Architecture and Extensibility](https://www.heyuan110.com/posts/ai/2026-09-14-pi-coding-agent-deep-dive/): An ultra-deep technical guide to Pi (pi.dev), the minimalist coding agent. Learn its dual-loop execution, branching session tree, local vs global packages, and how to build self-coding extensions.
- [Mastering Anthropic Agent SDK Hooks: Intercept & Secure](https://www.heyuan110.com/posts/ai/2026-09-14-mastering-anthropic-agent-hooks/): Learn how to use Anthropic's pretooluse and posttooluse hooks to secure tool-calling loops, prevent prompt injection, and optimize token costs with practical...
- [Stanford CS146S Fall 2026: Schedule, Videos, Follow Along](https://www.heyuan110.com/posts/ai/2026-09-11-cs146s-fall-2026-follow-along/): Stanford CS146S Fall 2026 runs Sep 22 to Dec 3. There are no official lecture videos: here is what you can actually watch, the full calendar, and a...
- [ego lite Review: Handing Claude Code Your Logged-In Browser](https://www.heyuan110.com/posts/ai/2026-09-10-ego-lite-browser-review/): ego lite v0.5 tested with Claude Code on 3 tasks vs agent-browser and Chrome DevTools MCP: 16s/$0.12 best, 195s worst. The token win is real; the...
- [Free AI Agent Courses Fall 2026: Stanford, CMU, MIT Compared](https://www.heyuan110.com/posts/ai/2026-09-09-free-ai-agent-courses-fall-2026/): Five free university AI agent courses for Fall 2026 compared: which have public lecture videos, which are materials-only, who each is for, and the stack to run.
- [CMU 11-768 AI Agents Fall 2026: Full Syllabus Breakdown](https://www.heyuan110.com/posts/ai/2026-09-08-cmu-11-768-ai-agents-course/): CMU 11-768 AI Agents (Fall 2026) by Graham Neubig and Daniel Fried: all 28 sessions, the harness, eval and RL assignments, grading, free videos, who should...
- [pi Coding Agent Review 2026: 4 Tools vs Claude Code, Tested](https://www.heyuan110.com/posts/ai/2026-09-06-pi-coding-agent-review/): pi coding agent vs Claude Code on 3 identical tasks: 1,358 vs 31,012 context tokens, 4 tools, zero permission dialogs. Real receipts and who should switch.
- [Stanford CS329Z Engineering AI Agents: Syllabus + Self-Study](https://www.heyuan110.com/posts/ai/2026-09-04-stanford-cs329z-engineering-ai-agents/): Stanford CS329Z (Fall 2026) teaches AI agents as engineering: decomposition, data, evals. Full dated schedule, HW1/HW2, grading, and a no-video self-study plan.
- [gitui vs lazygit in 2026: Benchmarked on 82K Commits](https://www.heyuan110.com/posts/ai/2026-09-02-gitui-vs-lazygit/): gitui vs lazygit on an 82K-commit repo: gitui paints in 11 ms with 31 MB, lazygit needs 416 ms but owns rebase, conflicts, worktrees and custom...
- [ubrowser Review: Fastest Cheapest Browser Automation? Tested](https://www.heyuan110.com/posts/ai/2026-07-27-ubrowser-review/): I installed lulzx/ubrowser and measured it against Playwright MCP: 761 vs 7,400 tokens on the same page. The design wins. The project is the problem.
- [Claude Code Screenshot MCP Setup: Browser Automation 2026](https://www.heyuan110.com/posts/ai/2026-07-21-claude-code-screenshot-mcp-frontend-debugging/): Set up screenshot MCP in Claude Code in one command. Chrome DevTools vs Playwright MCP for browser testing and frontend debugging, with measured token costs.
- [Codex App Guide (2026): Setup, Workflows, and Real Pitfalls](https://www.heyuan110.com/posts/ai/2026-07-12-openai-codex-app-guide/): The Codex app now lives inside the ChatGPT desktop app. How to set it up, run parallel agents, decode the credit pricing, and when the CLI is the better tool.
- [Seedance 2.0 Prompt Guide: Best Practices & Failure Modes](https://www.heyuan110.com/posts/ai/2026-07-11-seedance-2-prompt-guide/): How to write Seedance 2.0 prompts that work: a five-slot formula, @reference tagging rules, six failure modes, and what transfers to Veo 3 and Sora 2.
- [GPT-5.6 Release: Pricing, ChatGPT Work, and the Codex Merger](https://www.heyuan110.com/posts/ai/2026-07-10-gpt-5-6-general-availability/): GPT-5.6 went GA on July 9, 2026: Sol/Terra/Luna pricing, the Codex-ChatGPT Work merger, what OpenAI benchmark claims hide, and whether to switch from Claude.
- [Claude Fable 5: When the $10/$50 Flagship Is Worth It](https://www.heyuan110.com/posts/ai/2026-07-10-claude-fable-5-guide/): Fable 5 is usage-credits only after July 12, 2026. Real tasks cost $30-100, not the $10/$50 sticker. A decision tree for when it pays off, and how to prompt it.
- [Seedream 5.0 Pro: ByteDance's Image Model Takes On Gemini](https://www.heyuan110.com/posts/ai/2026-07-09-seedream-5-pro/): Seedream 5.0 Pro launched July 8, 2026 at $0.045 per 1K image. I fact-check the Gemini comparison, walk through the API, and map who should actually switch.
- [What SpaceX's $60B Cursor Acquisition Means for Developers](https://www.heyuan110.com/posts/ai/2026-07-09-spacex-cursor-acquisition/): SpaceX confirmed a $60B all-stock deal for Cursor maker Anysphere, closing Q3 2026. What it means for Claude model access, pricing, and whether to switch.
- [Lovable vs v0 vs Bolt: AI App Builders Compared 2026](https://www.heyuan110.com/posts/ai/2026-07-08-lovable-vs-v0-vs-bolt/): Lovable if you can't code, v0 if you ship Next.js on Vercel, Bolt for a browser IDE and Expo mobile — and when engineers should skip all three for Claude Code.
- [Claude AI Free Tier Limits 2026: Is Free Enough?](https://www.heyuan110.com/posts/ai/2026-07-08-claude-free-tier-limits/): Claude free plan 2026: how many messages you get (15-40 per 5 hours), what free can and cannot do, and the exact signals that mean you should upgrade to Pro.
- [Best AI Coding Models 2026: Fable 5 vs Sonnet 5 vs GPT-5.6](https://www.heyuan110.com/posts/ai/2026-07-07-best-ai-coding-models-2026/): Sonnet 5 is the best AI coding model for most developers in 2026 — it beats Opus 4.8 on Terminal-Bench at 40% of the price. Fable 5 is the escalation tier.
- [Loop Engineering: Building the Cage Your AI Agent Runs In](https://www.heyuan110.com/posts/ai/2026-07-05-loop-engineering/): Loop engineering is the 2026 discipline of building the cage your AI agent runs in: context hygiene, stop conditions, a real critic, and idempotent tools.
- [MCP vs Skills: Why CLI + Skill Wins the Agent Toolchain](https://www.heyuan110.com/posts/ai/2026-07-04-cli-skills-vs-mcp/): MCP servers burn 50K tokens where a 200-token SKILL.md does the same job. Why Perplexity, YC, and my own stack moved to CLI + Skills — and when MCP still wins.
- [Agentic Loops 2026: Self-Looping AI Agents Explained](https://www.heyuan110.com/posts/ai/2026-07-03-agentic-loops/): What agentic loops and the Ralph loop are, when self-looping AI agents ship code versus burn money, and the guardrails to run an autonomous coding agent in...
- [CS146S Study Guide 2026: Lecture-by-Lecture Notes & Workbook](https://www.heyuan110.com/posts/ai/2026-07-02-cs146s-study-guide/): Self-study Stanford CS146S in 2026: lecture-by-lecture verdicts, the exercises worth doing, Claude Code/Cursor tool mappings, and a route that skips filler.
- [OpenSpec vs Superpowers: My Spec-Driven Workflow](https://www.heyuan110.com/posts/ai/2026-06-28-openspec-superpowers-workflow/): How I run OpenSpec and Superpowers together in Claude Code: the CLAUDE.md routing that stops them fighting, plus a decision rule for when each pays off.
- [Best Windows Terminal 2026: Ranked for Developers](https://www.heyuan110.com/posts/ai/2026-06-22-best-windows-terminal-2026/): Windows Terminal + PowerShell 7 + WSL2 is the best Windows terminal for most developers in 2026. WezTerm for cross-platform, Warp for AI-first workflows.
- [Context Engineering for Coding Agents 2026: What Works](https://www.heyuan110.com/posts/ai/2026-06-16-context-engineering-2026/): Context engineering for coding agents in 2026: why subtracting context beats adding it — Sourcegraph data, offload-before-summarize thresholds, lean CLAUDE.md.
- [Anthropic's $965B Valuation Is Pricing Agent Infrastructure](https://www.heyuan110.com/posts/ai/2026-06-12-anthropic-965b-managed-agents/): Anthropic filed for an IPO at a $965B valuation on a $47B run-rate — a 20x multiple that prices Managed Agents, the agent runtime, not Claude itself.
- [Apple WWDC 2026: Siri Runs on Google Gemini — the Verdict](https://www.heyuan110.com/posts/ai/2026-06-09-apple-wwdc-2026-gemini-siri-pivot/): WWDC 2026 verdict: Siri now runs on Google Gemini and AFM Cloud Pro on NVIDIA GPUs in Google Cloud. Who won, who lost, why local AI on Macs comes out ahead.
- [Harness Engineering: Window of Opportunity, Not a Forever](https://www.heyuan110.com/posts/ai/2026-05-08-harness-engineering-window-of-opportunity/): Is Harness Engineering hype or substance? After 60 days in production, my call: a 2026-2027 window of opportunity, not a forever moat. Use this investment...
- [Hermes Agent v0.10 Review: How 113K Stars in 7 Weeks Hides](https://www.heyuan110.com/posts/ai/2026-04-24-hermes-agent-v010-deep-review/): Hermes Agent hit 113K GitHub stars 8 days after shipping v0.10 — the fastest open-source agent framework of 2026. We dissect the three-tier memory...
- [Managed Agents vs OpenClaw: Anthropic Just Enclosed the](https://www.heyuan110.com/posts/ai/2026-04-24-claude-managed-agents-vs-openclaw/): On April 4 Anthropic blocked OpenClaw. On April 16 they shipped Managed Agents beta. 12 days between killing the community harness and shipping the official...
- [wshobson/agents Audit: 184 Claude Code Agents, 78 Plugins](https://www.heyuan110.com/posts/ai/2026-04-20-wshobson-agents-deep-dive/): The 33.9K-star Claude Code marketplace audited piece by piece: 78 plugins, 184 agents, 150 skills, which 2-4 to pick, and why installing all is a trap.
- [Harness Engineering: Build the 6 Layers Backwards (6 to 1)](https://www.heyuan110.com/posts/ai/2026-04-18-harness-six-layers-reverse-build/): 60 days running an AI coding harness in production: layers 5 and 6 (eval and recovery) drive about 80% of stability. Build 6 to 1, not 1 to 6, with ROI numbers.
- [Playwright CLI + Skills: 0-Token Browser Automation Pattern](https://www.heyuan110.com/posts/ai/2026-04-18-playwright-cli-skill-zero-token-automation/): Browser automation costs collapse in 3 stages — explore with AI (41% context), freeze as a Skill (5%), ship as a script (0 tokens). Hands-on...
- [Claude Agent SDK in Python: query() vs ClaudeSDKClient](https://www.heyuan110.com/posts/ai/2026-04-17-claude-agent-sdk-guide/): query() opens a fresh session per call; ClaudeSDKClient keeps multi-turn context. Custom MCP tools, three-layer permissions and production patterns.
- [Hermes Agent Review 2026: Nous Research Setup + Best Models](https://www.heyuan110.com/posts/ai/2026-04-14-hermes-agent-guide/): Nous Research Hermes Agent v0.9.0 hands-on review (April 2026): 27K+ GitHub stars, installation guide on $5 Hetzner VPS, best models for the harness (Claude...
- [Mac Local LLM Benchmarks: M3 Max vs M4 Pro on Llama 70B](https://www.heyuan110.com/posts/ai/2026-04-14-mac-apple-silicon-ai-workstation/): Llama 3.3 70B runs at 9.8 tok/s on a 128GB M3 Max Studio, 7.5 on M3 Max 64GB, and OOMs on M4 Pro 48GB. Memory bandwidth sets tok/s; the Neural Engine idles.
- [OpenClaw Tools & Skills Reference 2026: Every Built-in Tool](https://www.heyuan110.com/posts/ai/2026-04-14-openclaw-tools-skills-reference/): Complete OpenClaw 2026 reference covering built-in tools (Read, Write, Edit, Bash, Grep, Glob, Task, TodoWrite) with exact write tool parameters (path...
- [Terminal AI Coding Tools 2026: Three Lanes, Not One Race](https://www.heyuan110.com/posts/ai/2026-04-14-terminal-ai-coding-tools-2026-comparison/): Deep comparison of Claude Code, Codex CLI, Gemini CLI, and Aider. Instead of a flat feature table, this guide maps the 2026 terminal AI landscape into three...
- [Sub-Agent Architecture for AI Coding Harnesses: When to](https://www.heyuan110.com/posts/ai/2026-04-13-harness-subagent-architecture/): Most teams use sub-agents as a speed hack. The real value is context garbage collection. A practical guide to sub-agent architecture, Opus/Sonnet/Haiku...
- [Lazygit Guide: Interactive Rebase in a Terminal Git TUI](https://www.heyuan110.com/posts/ai/2026-04-10-lazygit-guide/): Lazygit turns interactive rebase and line-level staging into single keystrokes. 76K stars, written in Go, plus how it stacks up against gitui (22K) and tig.
- [Claude HUD 2026: Claude Code's 18K-Star Status Bar in 3 Min](https://www.heyuan110.com/posts/ai/2026-04-10-claude-hud-guide/): Claude HUD is a free Claude Code plugin: a live status bar for context usage, rate limits, and agent activity. Install with one /plugin command in 3 minutes.
- [OpenSpec vs Superpowers in Claude Code: Who Owns What](https://www.heyuan110.com/posts/ai/2026-04-09-claude-code-openspec-superpowers/): OpenSpec and Superpowers fix different failures: Superpowers enforces branches, tests and review, while only OpenSpec Delta/Archive preserves design rationale.
- [One Founder + AI Agents = 2,000 New Customers/Month: The](https://www.heyuan110.com/posts/ai/2026-04-04-ai-agent-marketing-automation/): How Postiz founder Nevo David used Paperclip, Claude Code, and a Skill system to build an AI marketing department. The 6-layer stack, 3-step starter...
- [Claude Code Buddy: The Terminal Pet Hidden in Your AI](https://www.heyuan110.com/posts/ai/2026-04-04-claude-code-buddy-terminal-pet/): A deep dive into Claude Code Buddy — the deterministic terminal pet system with 18 species, 5 rarity tiers, and a dual-layer architecture that makes each...
- [Claude Pricing 2026: Pro $20, Max $100/$200, Team and API](https://www.heyuan110.com/posts/ai/2026-04-03-claude-pricing-complete-guide/): Claude Pro costs $20/mo, Max 5x $100, Max 20x $200, Team $25-30 per seat, Free $0. Usage multipliers, API rates, and which tier pays off for daily Claude Code.
- [5 AI Coding Tools Compared: Why Picking Just One Is the](https://www.heyuan110.com/posts/ai/2026-04-03-claude-code-vs-cursor-vs-copilot/): Claude Code, Cursor, Copilot, Codex CLI, and Gemini CLI after 8 months of daily use. The $30/month combo that beats the $200/month single tool. Real...
- [OpenClaw Multi-Agent Setup: Stop Letting LLMs Orchestrate](https://www.heyuan110.com/posts/ai/2026-04-02-openclaw-multi-agent-setup-guide/): Complete OpenClaw multi-agent configuration guide. 8-tier binding priority, Lobster deterministic pipelines vs sessions_send, Feishu integration, cost...
- [MCP vs Skills vs Hooks in Claude Code: Which Extension Do](https://www.heyuan110.com/posts/ai/2026-04-02-mcp-vs-skills-claude-code/): Deep technical comparison of Claude Code MCP, Skills, and Hooks — protocol design, context costs, trigger mechanisms, and a decision framework for...
- [Cursor Composer 2 Review: The Kimi K2.5 Controversy and](https://www.heyuan110.com/posts/ai/2026-04-01-cursor-composer-2-review/): Deep technical analysis of Cursor Composer 2, built on Moonshot AI Kimi K2.5. Benchmarks, pricing, the attribution scandal, and honest comparison with Claude...
- [Claude Code Open Source: The Rewrite That Hit 100K Stars in](https://www.heyuan110.com/posts/ai/2026-04-01-claude-code-open-source-agent/): Deep technical analysis of Claude Code Open Source — the clean-room Python and Rust rewrite of Claude Code agent harness architecture, born from the...
- [Harness Engineering #2: The 60-Line CLAUDE.md Rule (and Why](https://www.heyuan110.com/posts/ai/2026-03-31-harness-claudemd-guide/): I shipped a 90-line CLAUDE.md and watched my agent get dumber. The ETH Zurich study explains why: LLM-generated files drop success by 20%...
- [Harness Engineering 2026: The Model Is the Least Important](https://www.heyuan110.com/posts/ai/2026-03-30-harness-engineering-guide/): 60 days running an AI coding harness in production. Model upgrades got me ~5%. Harness rewrites got me 60%+ cost drop with higher quality. A...
- [Seedance 2.0 Review: ByteDance AI Video vs Sora 2 and Veo 3](https://www.heyuan110.com/posts/ai/2026-03-29-seedance-2-bytedance-ai-video/): Seedance 2.0 tops the Artificial Analysis text-to-video board at Elo 1,269, ahead of Sora 2 and Veo 3. Free in CapCut, roughly $0.14 per 15-second clip via API.
- [Lark CLI Complete Guide: Control Feishu with Terminal and](https://www.heyuan110.com/posts/ai/2026-03-29-lark-cli-guide/): Lark CLI is the official command-line tool for Feishu/Lark Open Platform, covering 11 business domains with 200+ commands and 19 AI Agent Skills...
- [Cloudflare Tunnel vs frp vs SSH: Expose Localhost Safely](https://www.heyuan110.com/posts/ai/2026-03-28-cloudflare-tunnel-guide/): Expose 127.0.0.1:8080 with no public IP. Cloudflare Tunnel is free with a Cloudflare-managed domain, frp carries raw TCP and UDP, SSH -R needs a $5-20 VPS.
- [QMD Local Search: Cut AI Agent Token Costs by 90%](https://www.heyuan110.com/posts/ai/2026-03-25-qmd-local-search-ai-agent-memory/): QMD, the local hybrid search engine from Shopify CEO Tobi Lutke, turns a 6,500-token memory load into roughly 200. BM25, vectors, LLM reranking, offline.
- [Chrome DevTools MCP Setup 2026: Fix "Opens New Window" +](https://www.heyuan110.com/posts/ai/2026-03-17-chrome-devtools-mcp-guide/): How to connect AI to your already-logged-in Chrome browser. Fix the new-window problem in 3 steps, configure remote debugging port 9222 on macOS, enable...
- [GitHub Agent HQ: Multi-Agent Development Guide for VS Code](https://www.heyuan110.com/posts/ai/2026-03-13-github-agent-hq-vscode-guide/): Set up GitHub Agent HQ to run Claude, Codex, and Copilot in VS Code. Covers setup, multi-agent workflows, practical use cases, and comparisons with CLI tools.
- [AI Code Security Tools Compared: Codex Security vs Claude](https://www.heyuan110.com/posts/ai/2026-03-13-ai-code-security-tools-compared/): Compare OpenAI Codex Security, Claude Code Security, and Snyk for AI-powered vulnerability scanning. Real results, pricing, and which tool fits your security...
- [GPT-5.4 vs Claude Opus 4.6: Complete Comparison for](https://www.heyuan110.com/posts/ai/2026-03-13-gpt-5-4-vs-claude-opus-comparison/): GPT-5.4 and Claude Opus 4.6 compared head-to-head on coding benchmarks, agent capabilities, pricing, and real-world developer workflows in 2026.
- [OpenCode Review: Open Source AI Coding Agent, 2 Weeks In](https://www.heyuan110.com/posts/ai/2026-03-13-opencode-ai-coding-agent-review/): Two weeks with OpenCode as my main agent: MIT licence, 121K stars, 75+ model providers, built-in LSP, client-server on port 4096. Can it replace Claude Code?
- [AI Development Methodologies Compared: From Vibe Coding to](https://www.heyuan110.com/posts/ai/2026-03-11-ai-development-methodologies-compared/): Deep comparison of 6 AI development methodologies — Vibe Coding, SDD, BMAD, Ralph Loop, and pragmatic hybrid workflows. With Martin Fowler critique and...
- [TypeScript vs Python in the AI Era: Which Language Should](https://www.heyuan110.com/posts/ai/2026-03-10-typescript-vs-python-ai-era/): TypeScript overtook Python on GitHub with 66% growth driven by AI tools. Compare both languages for AI-assisted coding, ML development, full-stack apps, and...
- [High-Frequency Commits: Ship 100+ Commits/Day Without Chaos](https://www.heyuan110.com/posts/ai/2026-03-10-high-frequency-commits/): Learn how to sustain 100+ daily commits without breaking your codebase. Covers atomic commits, conventional commits, CI testing, and progressive release...
- [AI Dev Environment Setup: Tools, Configs, and Dotfiles for](https://www.heyuan110.com/posts/ai/2026-03-10-ai-dev-environment-setup/): Set up the ultimate AI development environment with Claude Code, Cursor, terminal configs, and dotfiles. Complete guide with templates, shell aliases, and...
- [Context Engineering: The Most Underrated AI Development](https://www.heyuan110.com/posts/ai/2026-03-10-context-engineering-guide/): Master context engineering to build better AI systems. Learn the five dimensions, four failure modes, and practical frameworks that separate hobbyist...
- [Prototype to Production: AI App Deployment Checklist](https://www.heyuan110.com/posts/ai/2026-03-09-prototype-to-production/): A complete checklist for deploying AI applications from prototype to production, covering testing, security, scalability, observability, and automated ops.
- [MCP Security 2026: 30 CVEs in 60 Days — What Went Wrong](https://www.heyuan110.com/posts/ai/2026-03-10-mcp-security-2026/): A deep dive into 30 MCP vulnerabilities filed in 60 days. Covers attack patterns, CVE timeline, OWASP Agentic Top 10, security tools comparison, and a...
- [Why TypeScript Surged 66%: How AI Tools Are Reshaping](https://www.heyuan110.com/posts/ai/2026-03-10-typescript-ai-tools/): TypeScript became GitHub #1 language with 66% growth, driven by AI coding tools. Learn why static types help LLMs generate better code and how the...
- [Cursor Agent Mode Guide 2026: Rules, Plan Mode, Worktrees](https://www.heyuan110.com/posts/ai/2026-03-08-cursor-setup-guide/): Cursor Agent Mode from a fresh install: Shift+Tab Plan Mode, .cursor/rules files, parallel agents isolated by Git Worktrees, and Cloud Agents.
- [Best AI Coding Agents 2026: 7 Tools Compared and Tested](https://www.heyuan110.com/posts/ai/2026-03-10-ai-coding-agents-comparison-2026/): Claude Code, Cursor, Antigravity, Copilot, Kiro, Codex CLI and Windsurf compared on price, autonomy and multi-file edits, plus the $40/mo combo most devs pick.
- [Kiro Review: Amazon Spec-Driven AI IDE, Tested in 2026](https://www.heyuan110.com/posts/ai/2026-03-10-kiro-review/): Amazon Kiro turns a prompt into specs, a plan, then code. Code OSS fork with Agent Hooks, Steering Files, MCP, and the 13-hour AWS outage it was blamed for.
- [Build an AI Coding Agent from Scratch in Python (Complete](https://www.heyuan110.com/posts/ai/2026-03-07-build-ai-agent-python/): Learn how to build an AI agent in Python with agentic loops, function calling, and tool use. Step-by-step tutorial with complete runnable code in 250 lines.
- [Codex CLI Deep Dive: Setup, Sandbox Modes, and 20+ Tips](https://www.heyuan110.com/posts/ai/2026-03-10-codex-cli-deep-dive/): Codex CLI reads ~/.codex/config.toml, where approval_policy, sandbox_mode, trust_level decide what it can run. Model switching, session recovery, MCP, 20+ tips.
- [Google Antigravity After 30 Days: 3 Killer Features, 5](https://www.heyuan110.com/posts/ai/2026-03-10-google-antigravity-review/): I used Google Antigravity as my daily driver for 30 days against Cursor and Claude Code. Here are the 3 features that genuinely surprised me — and the 5...
- [OpenAI Symphony: From Issue Ticket to Pull Request Without](https://www.heyuan110.com/posts/ai/2026-03-05-openai-symphony-autonomous-coding/): OpenAI Symphony monitors your issue tracker, spawns Codex agents, and delivers verified PRs automatically. Architecture breakdown, WORKFLOW.md setup, and...
- [OpenClaw vs CrewAI vs AutoGen 2026: 6 AI Agent Frameworks](https://www.heyuan110.com/posts/ai/2026-03-05-openclaw-vs-ai-agents/): OpenClaw AI agent framework vs CrewAI vs AutoGPT vs AutoGen vs LangGraph vs Devin — 6 frameworks ranked in 2026. Real setup cost (free to $500/mo)...
- [OpenClaw Tavily Search: ClawHub Skill Integration Guide](https://www.heyuan110.com/posts/ai/2026-03-05-openclaw-tavily-integration/): The clawhub tavily-search skill gives OpenClaw agents web search in minutes. Step-by-step install via clawdhub, API key setup, per-agent config, and cost...
- [OpenClaw Pitfalls: 15 Automation Mistakes and Fixes](https://www.heyuan110.com/posts/ai/2026-03-05-openclaw-automation-pitfalls/): Avoid costly OpenClaw automation mistakes. Learn 15 common pitfalls in setup, agent behavior, cost control, and security — with real examples and fixes.
- [Claude Code Slash Commands 2026: Complete List + Custom](https://www.heyuan110.com/posts/ai/2026-03-05-claude-code-slash-commands/): Every Claude Code slash command, keyboard shortcut, and CLI flag in one reference. 40+ built-in commands, custom Skills commands, and environment variables...
- [Build MCP Servers in Python: Complete Step-by-Step Tutorial](https://www.heyuan110.com/posts/ai/2026-03-05-build-mcp-server-python/): Learn to build MCP servers in Python with FastMCP. Complete tutorial covering tools, resources, prompts, testing with MCP Inspector, and deployment.
- [OpenClaw Setup Guide: Install and Configure Your AI Agent](https://www.heyuan110.com/posts/ai/2026-03-05-openclaw-setup-guide/): Complete OpenClaw setup guide. Step-by-step installation, Telegram/WhatsApp config, AI model setup, Skills, multi-agent basics, and security best practices.
- [OpenClaw Multi-Agent Setup: Build AI Teams That Work](https://www.heyuan110.com/posts/ai/2026-03-05-openclaw-multi-agent-setup/): Step-by-step OpenClaw multi-agent setup guide. Configure agent teams, communication patterns, Tavily search, messaging bindings, and 4 collaboration patterns.
- [Claude Code vs GitHub Copilot 2026: Which AI Coding Tool](https://www.heyuan110.com/posts/ai/2026-03-05-claude-code-vs-copilot/): Claude Code vs GitHub Copilot after 3 months: Copilot wins autocomplete ($10/mo), Claude Code wins autonomous coding ($20-200/mo). Real examples and pricing...
- [How Claude Code Hooks Work: 17 Lifecycle Events Explained](https://www.heyuan110.com/posts/ai/2026-03-05-claude-code-hooks-guide/): What Claude Code hooks are and how they work: all 17 lifecycle events explained, hook types, matcher rules, and exit codes — a beginner-friendly reference.
- [CLAUDE.md Best Practices: Write Files That Actually Work](https://www.heyuan110.com/posts/ai/2026-03-05-claude-code-claudemd-best-practices/): How to write effective CLAUDE.md files for Claude Code. Practical patterns, examples for 7 project types, and CLAUDE.md vs .cursorrules vs AGENTS.md compared.
- [Best MCP Servers for Claude Code: 18 Tools You Need in 2026](https://www.heyuan110.com/posts/ai/2026-03-05-best-mcp-servers-claude-code/): Curated list of the best MCP servers for Claude Code in 2026. Install commands, key features, and practical use cases for 18 top MCP servers across databases...
- [Claude AI Pricing 2026: Every Plan — Free, Pro $20, Max](https://www.heyuan110.com/posts/ai/2026-02-25-claude-code-pricing/): Complete Claude AI pricing 2026: Free (limited), Pro $20/mo, Max 5x $100/mo, Max 20x $200/mo. Real usage data, API vs subscription math, and which plan is...
- [OpenClaw 2026.3.1: WebSocket Streaming, Agent Routing, and](https://www.heyuan110.com/posts/ai/2026-03-03-openclaw-2026-3-1-new-features/): Complete guide to OpenClaw 2026.3.1 new features. Covers WebSocket-first transport, agent routing CLI, external secrets management, Kubernetes health...
- [Tmux Complete Guide: From Basics to AI-Powered Multi-Agent](https://www.heyuan110.com/posts/ai/2026-03-03-tmux-guide-ai-development/): Master tmux from zero to advanced AI development workflows. Learn session management, pane splitting, configuration, and how to run parallel Claude...
- [Vibe Coding Explained: What It Is and How to Do It Right](https://www.heyuan110.com/posts/ai/2026-02-28-vibe-coding-explained/): What is vibe coding? The AI development approach coined by Andrej Karpathy. Learn the workflow, best tools, security risks, and best practices for 2026.
- [MCP Protocol Explained: The Universal Standard for AI Tools](https://www.heyuan110.com/posts/ai/2026-02-28-mcp-protocol-explained/): What is Model Context Protocol (MCP) and why it matters in 2026. Architecture, tools vs resources vs prompts, building servers, and MCP vs function calling.
- [Building MCP Servers with TypeScript: Zero to Deploy](https://www.heyuan110.com/posts/ai/2026-03-02-building-mcp-servers-typescript/): Step-by-step guide to building MCP servers with TypeScript. Learn to create tools, resources, and prompts using the MCP SDK v2, test with Claude Code, debug...
- [Claude vs ChatGPT vs Gemini: Best LLM for Coding in 2026](https://www.heyuan110.com/posts/ai/2026-03-02-claude-vs-chatgpt-vs-gemini/): Compare Claude Opus 4.6, GPT-5.2, and Gemini 2.5 Pro for coding tasks. Real benchmarks, pricing, context windows, and use-case recommendations to...
- [GitHub Copilot vs Claude Code vs Cursor: 2026 Comparison](https://www.heyuan110.com/posts/ai/2026-02-28-copilot-vs-claude-vs-cursor/): GitHub Copilot vs Claude Code vs Cursor compared across pricing, agent mode, code completion, and IDE support. Find the best AI coding tool for 2026.
- [Claude Code vs Cursor 2026: Which AI Coding Tool Wins?](https://www.heyuan110.com/posts/ai/2026-02-28-claude-code-vs-cursor/): Claude Code vs Cursor head-to-head comparison for 2026. Agent capabilities, pricing, code quality, and real-world performance tested across 5 dimensions.
- [Claude Code Guide 2026: Everything You Need to Know](https://www.heyuan110.com/posts/ai/2026-02-28-claude-code-complete-guide/): The complete Claude Code guide for 2026. Installation, CLAUDE.md, MCP servers, Hooks, Skills, Worktree, Teams, pricing, and advanced workflows — all in one...
- [RAG Pipeline Setup: Vector Database + LLM Integration Guide](https://www.heyuan110.com/posts/ai/2026-03-01-rag-pipeline-setup/): Complete RAG pipeline tutorial with vector database setup, embedding strategies, chunking methods, and Python code examples using OpenAI, Chroma, and Qdrant.
- [Claude Rate Limits 2026: Real Caps for Pro, Max 5x and 20x](https://www.heyuan110.com/posts/ai/2026-02-28-claude-rate-limits/): Pro allows roughly 10-45 messages per 5 hours, Max 5x 50-200, Max 20x 200-800, with weekly hour caps on top. How the 5-hour window resets, and 3 workarounds.
- [Claude Code MCP Setup: Connect AI to Any External Service](https://www.heyuan110.com/posts/ai/2026-02-28-claude-code-mcp-setup/): Complete guide to MCP servers with Claude Code — install community servers, build your own in TypeScript, and debug like a pro.
- [CLAUDE.md Guide: Give AI Perfect Project Context Every Time](https://www.heyuan110.com/posts/ai/2026-02-28-claude-code-claudemd-guide/): Master CLAUDE.md to give Claude Code perfect project context every session. Three-layer config, templates, and real examples included.
- [Claude Code Worktree: Run Multiple AI Tasks in Parallel](https://www.heyuan110.com/posts/ai/2026-02-28-claude-code-worktree-guide/): Run multiple Claude Code sessions simultaneously with Git worktree mode. Parallel development, auto-cleanup, and team workflows.
- [AI Agent Security: Protecting Automated Workflows in 2026](https://www.heyuan110.com/posts/ai/2026-02-27-ai-agent-security/): Learn how to secure AI agent workflows against prompt injection, tool poisoning, and MCP vulnerabilities. Covers OWASP Agentic Top 10, real CVEs, defense...
- [Claude Code Teammate Mode: Multi-Agent Setup with tmux](https://www.heyuan110.com/posts/ai/2026-02-28-claude-code-teams-guide/): Run several Claude agents at once with teammate-mode: split-pane in tmux or iTerm2, mailbox messaging between agents, worktree isolation and token cost control.
- [Claude Code Skills: 20 I Use Daily + 3 That Saved Me 10](https://www.heyuan110.com/posts/ai/2026-02-28-claude-code-skills-guide/): The 20 Claude Code Skills I actually load every day, plus the 3 SKILL.md files that cut 10 hours/week off my workflow — with the exact frontmatter, triggers...
- [Claude Code Hooks: PreToolUse, PostToolUse & settings.json](https://www.heyuan110.com/posts/ai/2026-02-28-claude-code-hooks-guide/): How PreToolUse and PostToolUse hooks work in .claude/settings.json: hook events, matchers, exit codes, and 12 production configs with common failure modes.
- [Multi-Agent Orchestration: 4 Patterns That Actually Work](https://www.heyuan110.com/posts/ai/2026-02-26-multi-agent-orchestration/): Learn 4 proven multi-agent orchestration patterns — Pipeline, Router, Orchestrator-Worker, and Evaluator-Optimizer — with real-world examples from...
- [10 Claude Code Mistakes Beginners Make (And How to Fix Them)](https://www.heyuan110.com/posts/ai/2026-02-25-claude-code-mistakes/): The 10 most common Claude Code mistakes and how to fix them. From skipping CLAUDE.md to overusing Opus — save tokens and time.
- [Claude Code Install Guide 2026: Setup in 10 Minutes](https://www.heyuan110.com/posts/ai/2026-02-25-claude-code-setup-guide/): One command installs Claude Code; Node.js is no longer required. Native installer for macOS, Linux and Windows, npm migration via claude install, 5 error fixes.
- [Build Your Own Claude Code from Scratch in Python (250](https://www.heyuan110.com/posts/ai/2026-02-24-build-magic-code/): A hands-on tutorial that demystifies Claude Code by rebuilding its core architecture — the Agentic Loop, Tool Use, and streaming — from scratch in Python. Go...
- [Stanford CS146S Deep Dive (5): From Prototype to Production](https://www.heyuan110.com/posts/ai/2026-02-24-prototype-to-production/): A deep analysis of Stanford CS146S Weeks 8-9: building an app with one prompt is just the beginning. Learn the complete path from demo to production...
- [Stanford CS146S Deep Dive (4): Secure Vibe Coding — AI Code](https://www.heyuan110.com/posts/ai/2026-02-24-secure-vibe-coding/): Deep dive into Stanford CS146S Weeks 6-7: real-world Prompt Injection to RCE exploit, OWASP Top 10 in the Agent era, AI code review methodology, and how to...
- [Stanford CS146S Deep Dive (3): Agent Manager — Best](https://www.heyuan110.com/posts/ai/2026-02-24-agent-manager-patterns/): A deep dive into Stanford CS146S Week 4: from Claude Code creator Boris Cherney's talk to Anthropic's official best practices, learn how to manage Coding...
- [Context Engineering: The Most Underrated Core Skill in AI](https://www.heyuan110.com/posts/ai/2026-02-24-context-engineering-deep-dive/): Deep dive into Stanford CS146S Week 3: the paradigm shift from Prompt Engineering to Context Engineering, why Specs are the new source code, and four failure...
- [Stanford CS146S: The Modern Software Developer Syllabus](https://www.heyuan110.com/posts/ai/2026-02-24-stanford-cs146s-overview/): Stanford CS146S covers AI-driven development in 10 weeks, LLM basics to agent security. Slides, readings and code are free; Fall 2026 runs Sep 22 to Dec 3.
- [OpenClaw Multi-Agent Guide: Architecture, Configuration,](https://www.heyuan110.com/posts/ai/2026-02-23-openclaw-multi-agent-guide/): A deep dive into multi-agent design inside the OpenClaw AI agent framework (2026). From single-agent bottlenecks to building agent teams, covering routing...
- [2026 Agentic Coding Trends: 8 Key Insights Behind Claude](https://www.heyuan110.com/posts/ai/2026-02-23-agentic-coding-trends-2026/): Deep dive into Anthropic 2026 Agentic Coding trends report: from Claude Code hitting $1B in 6 months to Anthropic $14B ARR, breaking down 8 major AI coding...
- [MCP Security Guide: Attack Patterns, Real CVEs, and Defense](https://www.heyuan110.com/posts/ai/2026-02-23-mcp-security-guide/): Complete MCP security analysis covering OWASP Agentic Top 10, real CVE cases, mcp-scan tooling, and practical defense strategies for securing AI agent...
- [Build an MCP Server with Claude Code: TypeScript Tutorial](https://www.heyuan110.com/posts/ai/2026-02-22-claude-code-mcp-server-tutorial/): Learn how to build a custom MCP Server using Claude Code and TypeScript. Covers project setup, tool registration, debugging, and publishing to npm with full...
- [Claude Code Security: How AI-Powered Code Scanning Changes](https://www.heyuan110.com/posts/ai/2026-02-22-claude-code-security/): Deep dive into Anthropic Claude Code Security: how AI vulnerability scanning works, real-world results with 500+ bugs found, comparison with traditional SAST...
- [Claude Code February 2026 Updates: Worktree, Background](https://www.heyuan110.com/posts/ai/2026-02-22-claude-code-february-updates/): Claude Code shipped major features in February 2026 including Git Worktree parallel development, background agent tasks, and a leaner Simple Mode. Full...
- [Vibe Coding Guide 2026: What It Is, Best Tools, and How to](https://www.heyuan110.com/posts/ai/2026-02-22-vibe-coding-guide/): Learn vibe coding from basics to advanced. Compare top tools (Claude Code, Cursor, Trae), see real workflow examples, and discover why 91% of...
- [Claude Code Agent Teams 2026: Parallel Multi-Agent](https://www.heyuan110.com/posts/ai/2026-02-22-claude-code-agent-teams/): Complete guide to Claude Code Agent Teams: run multiple Claude instances in parallel, Opus+Sonnet mixed orchestration for cost savings, real-world workflows...
- [Agent Memory vs RAG: Context Engineering for LLM Agents](https://www.heyuan110.com/posts/ai/2026-02-21-ai-agent-memory-systems/): RAG retrieves at query time; context engineering curates the window upfront. Why a 200K window dies after 50 tool calls, plus a 4-layer memory stack.
- [Claude Code Worktree: Run Multiple AI Tasks in One Repo](https://www.heyuan110.com/posts/ai/2026-02-20-claude-code-worktree/): Learn how to use Claude Code --worktree (-w) mode to run parallel AI coding sessions in isolated directories. Covers setup, auto-cleanup, best practices, and...
- [MCP Protocol Explained: The Universal Standard for AI](https://www.heyuan110.com/posts/ai/2026-02-20-mcp-protocol-guide/): Complete guide to Model Context Protocol (MCP) — architecture, core primitives, MCP Apps interactive UI, Linux Foundation governance, Function Calling...
- [Xcode 26.3 Agentic Coding: Claude Agent & Codex in Apple IDE](https://www.heyuan110.com/posts/ai/2026-02-20-xcode-agentic-coding/): Xcode 26.3 integrates Anthropic Claude Agent and OpenAI Codex with full agentic coding capabilities. Deep dive into setup, features, MCP support, and how it...
- [Figma Code to Canvas: Turn AI-Generated Code into Editable](https://www.heyuan110.com/posts/ai/2026-02-19-figma-code-to-canvas/): Figma and Anthropic launch Code to Canvas, converting Claude Code UI output into editable Figma frames via MCP. Complete setup guide, use cases, and...
- [Claude Code vs Codex CLI (2026): 8-Dimension Head-to-Head](https://www.heyuan110.com/posts/ai/2026-02-19-claude-code-vs-codex/): Claude Code Opus 4.6 vs ChatGPT Codex GPT-5.3 compared across coding quality, context window, agent collaboration, pricing, and more. Real-world testing to...
- [Claude Code vs Cursor vs Windsurf 2026: Speed, Cost &](https://www.heyuan110.com/posts/ai/2026-02-18-claude-code-vs-cursor-vs-windsurf-2026/): Hands-on comparison of Claude Code, Cursor, and Windsurf in 2026 covering speed, cost, controllability, and learning curve with actionable selection...
- [Claude Code Hooks Examples: 12 Copy-Paste Automation Configs](https://www.heyuan110.com/posts/ai/2026-02-18-claude-code-hooks-guide/): 12 copy-paste Claude Code hook configs: auto-format on save, protect .env files, block rm -rf, Git auto-stage, Slack alerts, plus a full team setup example.
- [MoltBot Explained: What It Is, How It Works, and Why It](https://www.heyuan110.com/posts/ai/2026-02-18-what-is-moltbot/): MoltBot (now OpenClaw) is an open-source personal AI agent that runs locally and executes real tasks via Telegram and WhatsApp. Learn its architecture...
- [Draw Things API + MCP: Local Image Generation in Claude Code](https://www.heyuan110.com/posts/ai/2026-02-16-claude-code-draw-things-workflow/): Draw Things exposes an HTTP API on localhost:7860; mcp-drawthings bridges it to Claude Code over MCP. Setup in 3 steps, 4 core tools, free and fully offline.
- [OpenClaw 30-Day Rise: 180K Stars, 40+ Vulnerabilities,](https://www.heyuan110.com/posts/ai/2026-02-16-openclaw-openai-analysis/): How the OpenClaw AI agent framework went from a solo open-source project to 180K GitHub stars in 2026, exposed 40+ security flaws, spawned the first...
- [Draw Things on Mac 2026: Tutorial + 40% Faster than ComfyUI](https://www.heyuan110.com/posts/ai/2026-02-15-draw-things-ultimate-guide/): Draw Things is a free Mac/iOS app for local AI image generation, 20-40% faster than ComfyUI on Apple Silicon. Setup, LoRA training, Flux, and Claude Code MCP.
- [Draw Things vs ComfyUI vs DiffusionBee on Mac Mini M4](https://www.heyuan110.com/posts/ai/2026-02-15-mac-mini-local-image-generation/): Draw Things renders Flux 1024x1024 in 50s on a Mac Mini M4 Pro 24GB, about 20% faster than ComfyUI. DiffusionBee has not shipped a release since August 2024.
- [OpenClaw Architecture Deep Dive: How Automation Actually](https://www.heyuan110.com/posts/ai/2026-02-14-openclaw-architecture-deep-dive/): A complete walkthrough of the OpenClaw AI agent framework internals (2026) — how Gateway routes messages to Agents, how Skills orchestrate tools, how Nodes...
- [OpenClaw Automation Pitfalls: Installing 3 Skills Is Not](https://www.heyuan110.com/posts/ai/2026-02-14-openclaw-automation-pitfalls/): Installing tavily-search, find-skills, and proactive-agent on OpenClaw raises the capability ceiling, but without session isolation, task scheduling, and...
- [Codex CLI Mastery Guide: 20+ Power Tips for Real-World](https://www.heyuan110.com/posts/ai/2026-02-12-codex-cli-mastery-guide/): Complete Codex CLI guide covering installation, model switching, session recovery, MCP integration, security modes, and an honest Codex CLI vs Claude Code...
- [OpenClaw Tutorial: Complete Setup Guide for Beginners and](https://www.heyuan110.com/posts/ai/2026-02-12-openclaw-usage-tutorial/): Step-by-step OpenClaw tutorial covering installation, Gateway setup, TUI terminal, multi-agent configuration, session isolation, and troubleshooting for your...
- [Claude-Mem Deep Dive: Persistent Memory Plugin for Claude](https://www.heyuan110.com/posts/ai/2026-02-03-claude-mem-deep-dive/): Claude-Mem gives Claude Code cross-session persistent memory via hooks, AI compression, hybrid search, and Endless Mode. Full architecture breakdown and...
- [Moltbook Explained: The AI-Only Social Network With 150K](https://www.heyuan110.com/posts/ai/2026-02-01-moltbook-ai-agent-social-network/): Moltbook is a social network exclusively for AI agents where 150,000 bots created religions, debated consciousness, and formed governments. Full setup guide...
- [Superpowers Deep Dive: The Skills Framework That Makes](https://www.heyuan110.com/posts/ai/2026-02-01-superpowers-deep-dive/): Superpowers is a 40K-star agentic skills framework for Claude Code. It enforces TDD, subagent-driven development, and structured planning to transform AI...
- [OpenClaw Claude Code Workflow: How One Dev Built a](https://www.heyuan110.com/posts/ai/2026-01-31-openclaw-claude-code-workflow/): Deep dive into Peter Steinberger OpenClaw development methodology using Claude Code. Learn AGENTS.md documentation-driven development, multi-agent parallel...
- [22 Thinking Frameworks That Turn Vague Ideas Into Clear](https://www.heyuan110.com/posts/ai/2026-01-31-thinking-methodologies-guide/): A practical guide to SMART, OKR, MECE, JTBD, RICE, PDCA, and 16 more proven methodologies for defining goals, breaking down problems, writing requirements...
- [Unit Test Report Tools: Framework Comparison, Coverage](https://www.heyuan110.com/posts/ai/2026-01-31-unit-test-report-tools/): Compare top unit test reporting tools in 2026 including JUnit, PyTest, Jest, Allure Report, and AI-assisted testing workflows to build effective test...
- [CLAUDE.md vs README.md: What Goes in Each File](https://www.heyuan110.com/posts/ai/2026-01-31-claudemd-vs-readme/): README.md describes the project to humans; CLAUDE.md tells agents what to do. The exact split of rules and commands, plus why 300 lines is the ceiling.
- [High-Frequency Commits: Engineering Practices for 100+](https://www.heyuan110.com/posts/ai/2026-01-31-high-frequency-commits-strategy/): High-frequency commits do not mean chaos. Learn atomic commits, Conventional Commits, layered testing, progressive rollouts, and a 10-point checklist to ship...
- [OpenClaw Memory Strategy: Tool-Driven RAG and On-Demand](https://www.heyuan110.com/posts/ai/2026-01-31-openclaw-memory-strategy/): How OpenClaw implements agent memory using tool-driven RAG instead of prompt injection — combining BM25 and vector search, chunk-based indexing, and...
- [Delete Before You Optimize: 3 Product Thinking Rules for](https://www.heyuan110.com/posts/ai/2026-01-30-product-thinking-delete-first/): A practical product methodology for indie hackers and solopreneurs: delete before optimizing, question every assumption, and observe real users. With...
- [AI Workflow Playbook: From Prompts to Production Code](https://www.heyuan110.com/posts/ai/2026-01-30-ai-workflow-real-guide/): A battle-tested AI workflow methodology covering prompt engineering, AI-assisted coding, tool selection, and enterprise deployment — turn AI from a chat toy...
- [Moltbot Explained: 80K Stars, Renaming Drama & Security](https://www.heyuan110.com/posts/ai/2026-01-29-moltbot-deep-dive/): Moltbot (formerly Clawdbot/OpenClaw) is the fastest-growing open-source AI Agent of 2026. Learn about its architecture, the Anthropic-forced rename, crypto...
- [Browser Automation in Claude Code: 5 Tools Compared (2026)](https://www.heyuan110.com/posts/ai/2026-01-28-claude-code-browser-automation/): Browser-use vs Agent Browser vs Playwright CLI vs Playwright MCP vs DevTools MCP — real-world token usage differs by 10x+. Speed, cost, and stability...
- [Moltbot Wizard Setup Guide: Build Your Own Private AI](https://www.heyuan110.com/posts/ai/2026-01-28-moltbot-wizard-guide/): Step-by-step guide to configuring Moltbot using the Wizard CLI. Set up Gateway, connect Telegram and WhatsApp, install Skills, and deploy your personal AI...
- [AGI Is Here in 2026: Sequoia Capital Case Study and What It](https://www.heyuan110.com/posts/ai/2026-01-26-agi-is-here/): Sequoia Capital declares AGI has arrived. See how an AI agent completed a full recruiting job in 31 minutes, why agent capabilities double every 7...
- [Claude Code + Remotion: Generate Pro Videos with AI](https://www.heyuan110.com/posts/ai/2026-01-26-claude-code-remotion-video/): Learn how to create professional animated videos using Claude Code and Remotion. This step-by-step guide covers setup, workflow, and real-world examples of...
- [ClawdBot Setup Guide: Build Your Personal AI Assistant in](https://www.heyuan110.com/posts/ai/2026-01-25-clawdbot-personal-ai-assistant/): Learn how to set up ClawdBot, an open-source AI agent that runs 24/7 on your computer. This step-by-step guide covers installation, Telegram integration, and...
- [Taste Matters More Than Ever in the AI Era: Your Only Moat](https://www.heyuan110.com/posts/ai/2026-01-23-taste-matters-in-ai-era/): When AI can generate infinite content instantly, the real bottleneck is no longer creation but judgment. Learn why personal taste is the scarcest...
- [Claude Code: 24 Power Tips to Master the AI Terminal](https://www.heyuan110.com/posts/ai/2025-01-23-claude-code-commands/): Master Claude Code with 24 practical tips covering shortcuts, CLAUDE.md config, MCP integration, custom slash commands, session management, and cost control...
- [Anthropic CEO Dario Amodei Bloomberg Interview: Key](https://www.heyuan110.com/posts/ai/2025-01-21-anthropic-ceo-bloomberg-interview/): Anthropic CEO Dario Amodei shares insights on AI exponential growth, chip export controls, superintelligence, economic disruption, and why his lead engineer...
- [Claude Code Skills: Top 20 Most Popular Skills in 2026](https://www.heyuan110.com/posts/ai/2026-01-20-claude-code-skills-top20/): Discover the 20 most popular Claude Code Skills ranked by GitHub stars and usage. Covers dev workflows, AI/LLM development, specialized tools, and skill...
- [AI Development Workflow: From Requirements to Production](https://www.heyuan110.com/posts/ai/2026-01-19-ai-dev-workflow/): A practical guide to integrating AI tools like Claude Code and Cursor into every stage of software development — from requirements analysis and architecture...
- [Agent Skills: Why Markdown Files Are the New Programs](https://www.heyuan110.com/posts/ai/2026-01-19-agent-skills-new-programming/): Agent Skills turn plain Markdown into executable programs that anyone can write, modify, and share. Discover how natural language programming is reshaping...
- [Cursor Agent Best Practices: Plan Mode, Rules, Past Chats](https://www.heyuan110.com/posts/ai/2026-01-19-cursor-agent-best-practices/): Plan first with Shift+Tab Plan Mode and keep plans in .cursor/plans/. Pull context via @Past Chats, split static Rules from dynamic Skills, add a stop hook.
- [Claude Code Complete Guide: From Beginner to Power User](https://www.heyuan110.com/posts/ai/2026-01-14-claude-code-guide/): Master Claude Code, Anthropic official CLI tool. Learn installation, CLAUDE.md setup, Skills, Subagents, MCP integration, and advanced workflows for...
- [Claude Cowork: Anthropic Desktop AI Agent That Controls](https://www.heyuan110.com/posts/ai/2026-01-13-claude-cowork/): Claude Cowork is a research preview desktop AI agent that directly accesses local files, runs code in sandboxed environments, and integrates with Notion...
- [Vercel Agent Browser: AI-Native Browser Automation CLI Tool](https://www.heyuan110.com/posts/ai/2026-01-13-vercel-agent-browser/): Vercel open-sourced Agent Browser: a Rust CLI plus Node.js daemon that hands agents an accessibility snapshot with refs like @e1, not CSS selectors.
- [Claude Code Skills: Create Custom AI Abilities in 30 Seconds](https://www.heyuan110.com/posts/ai/2026-01-12-claudecode-skill-patterns/): Learn how to build Claude Code Skills that turn your expertise into reusable AI modules. Step-by-step guide with practical SKILL.md examples for...
- [CLAUDE.md Guide: Give Claude Code Persistent Memory](https://www.heyuan110.com/posts/ai/2026-01-12-claudemd-memory-guide/): Learn how to configure CLAUDE.md so Claude Code remembers your tech stack, coding standards, and workflow preferences across every conversation.
- [Claude Code Skills Guide: Teach AI Your Exact Workflow](https://www.heyuan110.com/posts/ai/2026-01-08-claudecode-skill-guide/): Learn how to create Claude Code Skills with three practical examples: financial reporting, interview question generation, and Amazon listing optimization.
- [Claude Code Best Practices: 5 Tips from the Founder](https://www.heyuan110.com/posts/ai/2026-01-06-claudecode-best-practices/): Practical Claude Code tips shared by its founder — parallel agents, model selection, CLAUDE.md maintenance, slash commands, and verification loops.
- [Skills vs MCP in Claude Code: Two Ways to Extend AI](https://www.heyuan110.com/posts/ai/2026-01-06-skillmcp/): Understand the key differences between Skills and MCP (Model Context Protocol) in Claude Code — when to use each, how they manage context, and how they work...
- [Claude Code Skills vs SubAgents: Context Management Guide](https://www.heyuan110.com/posts/ai/2025-12-26-claudecode-skillsubagent/): Learn the key differences between Skills and SubAgents in Claude Code. Understand when to internalize capabilities vs delegate to child agents for optimal...
- [Vector Database Explained: From Core Concepts to Production](https://www.heyuan110.com/posts/ai/2025-03-11-vectordatabase/): Learn what vector databases actually solve, how ANN indexing works, when to use semantic search vs keyword search, and how to choose between Milvus, Pinecone...
- [FaceFusion Guide: Open-Source AI Face Swap Setup and Tips](https://www.heyuan110.com/posts/ai/2024-10-16-facefusion/): Complete FaceFusion tutorial covering installation, GPU requirements, parameter tuning, and troubleshooting for the best open-source AI face swapping tool.

## DATAWAREHOUSE — 1


- [Amazon Redshift Performance Tuning: VACUUM, ANALYZE, and](https://www.heyuan110.com/posts/datawarehouse/2018-08-09-dw-redshift/): A complete guide to Amazon Redshift performance optimization covering all 6 VACUUM types (FULL, DELETE ONLY, SORT ONLY, REINDEX, RECLUSTER, BOOST), ANALYZE...

## DOCKER — 6


- [Docker Compose Tutorial: docker-compose.yml Explained with](https://www.heyuan110.com/posts/docker/2026-01-24-docker-compose-yml-explained/): Complete Docker Compose tutorial covering every docker-compose.yml field — services, volumes, networks, healthcheck, ports — with a WordPress + MySQL...
- [Cloudflare Workers Complete Guide: Edge Computing from](https://www.heyuan110.com/posts/docker/2026-01-23-cloudflare-workers-guide/): Learn how to build and deploy serverless applications with Cloudflare Workers. Covers core concepts, free tier limits, use cases, development workflow, and...
- [Docker Compose Complete Guide (2026): Install, Configure,](https://www.heyuan110.com/posts/docker/2026-01-19-docker-compose-complete-guide/): Master Docker and Docker Compose from scratch: installation, core concepts, docker-compose.yml deep dive, WordPress and Node.js deployment examples, plus...
- [Docker Commands Cheat Sheet: Images, Containers, Networks,](https://www.heyuan110.com/posts/docker/2019-11-14-docker-commands/): A practical Docker command reference covering image management, container lifecycle, port mapping, volume mounting, and ready-to-use run commands for MySQL...
- [Nexus3 Private Docker Registry Setup: Enterprise Container](https://www.heyuan110.com/posts/docker/2019-06-12-next3-dockerhub/): Step-by-step guide to setting up an enterprise Docker private registry with Nexus3, covering hosted, proxy, and group repository types, image push/pull...
- [Docker Beginner Tutorial: Images, Containers, Registries,](https://www.heyuan110.com/posts/docker/2019-05-13-learn-docker/): A comprehensive Docker tutorial for beginners covering the three core concepts — images, containers, and registries — along with layered storage, Dockerfile...

## ELASTICSEARCH — 3


- [AWS EKK Log System Setup: Elasticsearch + Kinesis + Kibana](https://www.heyuan110.com/posts/elasticsearch/2018-09-12-log-ekk/): Step-by-step tutorial for building an EKK log collection system on AWS using Amazon Elasticsearch Service, Kinesis, and Kibana to collect and analyze Nginx...
- [Elasticsearch Tutorial: Core Concepts of Indices,](https://www.heyuan110.com/posts/elasticsearch/2018-09-12-elasticsearch/): A beginner-friendly Elasticsearch tutorial covering core concepts like indices, types, and documents (with RDBMS analogies), plus hands-on examples of _cat...
- [ELK Stack Setup Guide: Elasticsearch + Logstash + Kibana +](https://www.heyuan110.com/posts/elasticsearch/2018-09-11-log-elk/): Complete guide to building an enterprise-grade ELK log collection and analysis system using Elasticsearch, Logstash, Kibana, and Kafka with Filebeat...

## GO — 1


- [Go Language Tutorial: Fundamentals of Golang Syntax and Data Types](https://www.heyuan110.com/posts/go/talk/2019-01-25-learn-go/): A beginner-friendly Go (Golang) tutorial covering language features like easy deployment, built-in concurrency, and strong performance. Learn Go program structure, basic syntax, data types, variables, constants, operators, control flow, functions, pointers, structs, slices, maps, interfaces, error handling, goroutines, and channels.

## JAVA — 2


- [Java Tutorial Part 2: Arrays Utility, Lambda Expressions,](https://www.heyuan110.com/posts/java/2024-05-13-learn-java2/): A hands-on Java tutorial covering the Arrays utility class, Lambda expressions, method references, regular expressions, exception handling, and...
- [Java Beginner Tutorial Part 1: JDK, JRE, JVM Explained and](https://www.heyuan110.com/posts/java/2024-05-05-learn-java/): A comprehensive Java beginner guide covering the Java ecosystem (SE/EE/ME), the differences between JDK, JRE, and JVM, IntelliJ IDEA setup, core syntax, OOP...

## LINUX — 14


- [Fish Shell 4.6 Review: Best Interactive Shell, Wrong Default](https://www.heyuan110.com/posts/linux/2026-04-18-fish-shell-rust-2026/): Fish shell 4.6 is the most polished interactive shell in 2026, but the Rust rewrite did not fix POSIX. Here is why I run fish as a front-end and keep zsh for...
- [Turn Your Office Mac Mini into a Personal VPN: 5 Approaches](https://www.heyuan110.com/posts/linux/2026-04-11-remote-access-mac-mini-vpn/): Five practical ways to remotely access your office Mac Mini from anywhere — Tailscale, WireGuard, ZeroTier, Cloudflare Tunnel, and frp. Full setup guides...
- [Bash Special Variables Explained: $$, $?, $@, $# and More](https://www.heyuan110.com/posts/linux/2019-05-13-linux-shell-vars/): A complete guide to Bash special variables including $$, $!, $?, $-, $*, $@, $#, and $0. Learn the difference between $* and $@, exit status codes, and best...
- [AWS CLI Complete Guide: Installation, Configuration, S3/EC2](https://www.heyuan110.com/posts/linux/2020-07-04-aws-cli/): Master AWS CLI from scratch: install v2, configure profiles, manage S3 buckets and EC2 instances, and troubleshoot common permission errors.
- [Nginx HTTPS Configuration Guide: Self-Signed Certs, Mixed Sites, and Reverse Proxy Pitfalls](https://www.heyuan110.com/posts/linux/nginx/nginx-https-complete-guide/): Practical guide to Nginx HTTPS covering self-signed SSL certificates with OpenSSL, selective HTTPS for mixed HTTP/HTTPS sites, and fixing DNS caching issues when reverse proxying to AWS ELB
- [curl Command Guide (2026): HTTP Requests, File Transfers,](https://www.heyuan110.com/posts/linux/2020-06-29-curl/): Complete curl tutorial covering GET/POST/PUT/DELETE requests, JSON payloads, file uploads and downloads, authentication, proxy setup, and performance timing.
- [Traceroute Command Explained: Network Path Tracing and](https://www.heyuan110.com/posts/linux/2020-06-28-traceroute/): Complete traceroute/tracert tutorial covering TTL mechanics, command options, output interpretation, and real-world troubleshooting on Linux, macOS, and...
- [Linux and macOS Command Cheat Sheet for Developers](https://www.heyuan110.com/posts/linux/2020-03-19-linux-mac-commands/): Essential Linux and macOS command reference covering networking, process management, file operations, text processing, disk management, and...
- [Setting Up Jira and Confluence on Ubuntu: Installation,](https://www.heyuan110.com/posts/linux/2019-04-15-jira-confluence-install/): Complete guide to installing Jira and Confluence on Ubuntu 16.04 with MySQL, including database setup, license activation, user directory integration...
- [Kerberods Cryptominer on Linux: Incident Response and](https://www.heyuan110.com/posts/linux/2019-04-12-linux-kerberods/): Real-world incident report of a Linux server compromised by the kerberods cryptomining malware — covering detection, investigation of mysterious CPU behavior...
- [Jenkins + AWS CodeDeploy + Auto Scaling: A Complete CI/CD](https://www.heyuan110.com/posts/linux/2018-11-20-jenkins-codedeploy-autoscaling/): Step-by-step guide to building a CI/CD pipeline with Jenkins, AWS CodeDeploy, and Auto Scaling — covering in-place and blue/green deployments, AppSpec...
- [Supervisor Linux Guide: Install, Config, systemd Compared](https://www.heyuan110.com/posts/linux/2018-10-07-supervisor/): Install Supervisor from source, write a [program] block with autorestart, drive it with supervisorctl, boot it from a systemd unit, and fix 4 common errors.
- [IP Addresses and CIDR Explained: A Complete Networking Guide](https://www.heyuan110.com/posts/linux/2018-10-06-ip-cidr/): Learn IPv4 address structure, classful addressing (A/B/C/D/E), subnet masks, and CIDR notation. Includes a subnet cheat sheet and practical Linux...
- [Oh My Zsh Setup Guide: Build a Productive Terminal](https://www.heyuan110.com/posts/linux/2015-06-17-shell-zsh/): Complete Oh My Zsh installation and configuration guide covering Zsh setup, Powerlevel10k theme, zsh-autosuggestions, zsh-syntax-highlighting plugins, custom...

## MACOS — 2


- [Terminal Emulator Comparison: 23 Tools Across 3 Platforms](https://www.heyuan110.com/posts/macos/2025-01-22-terminal-tools-guide/): 23 terminal emulators compared for macOS, Windows and Linux: iTerm2, Warp, Ghostty, Windows Terminal, Tabby, WezTerm, Alacritty, plus SSH and AI terminals.
- [Charles Proxy Tutorial: HTTP/HTTPS Interception, Mock Data,](https://www.heyuan110.com/posts/macos/charles/): A practical guide to Charles Proxy on macOS — learn how to intercept HTTP/HTTPS traffic, mock API responses, throttle network speed, use breakpoints, and...

## MANAGEMENT — 1


- [Ten Leadership Disciplines: Lessons on Responsibility,](https://www.heyuan110.com/posts/management/2019-08-19-liulan-management/): Key takeaways from a leadership training course covering the ten essential disciplines — from taking ownership and embracing failure to systems thinking and...

## MIDDLEWARE — 3


- [Redis Complete Guide: Installation, Data Types,](https://www.heyuan110.com/posts/middleware/2026-01-22-redis-complete-guide/): A comprehensive Redis guide covering installation on macOS, Windows, and Docker, all five core data types with commands, RDB/AOF persistence, replication...
- [RabbitMQ Tutorial: AMQP Protocol, Exchanges, and Message](https://www.heyuan110.com/posts/middleware/2019-08-02-mq-rabbitmq/): A comprehensive RabbitMQ tutorial covering the AMQP protocol, exchange types (direct, topic, fanout, headers), queue patterns, and practical PHP examples...
- [Message Queue Fundamentals: Core Concepts, Use Cases, and](https://www.heyuan110.com/posts/middleware/2019-07-31-mq/): A comprehensive guide to message queues in distributed systems — covering the Producer-Broker-Consumer model, key benefits (decoupling, async...

## MYSQL — 2


- [MySQL SQL Complete Guide: From Beginner to Advanced (With](https://www.heyuan110.com/posts/mysql/2026-01-23-mysql-sql-complete-guide/): Master MySQL SQL from basics to advanced topics: indexes, B+ trees, transactions, MVCC, locks, and query optimization. Includes real-world examples...
- [MySQL EXPLAIN: Read Execution Plans and Optimize Queries](https://www.heyuan110.com/posts/mysql/2019-09-06-mysql-explain/): All 12 MySQL EXPLAIN fields decoded, with type and Extra covered in depth, plus FORMAT=JSON query cost, EXPLAIN ANALYZE, and 4 worked optimization examples.

## PYTHON — 3


- [uv vs conda vs pyenv: Which Python Manager Wins in 2026?](https://www.heyuan110.com/posts/python/2026-07-27-uv-vs-conda-vs-pyenv/): uv vs conda vs pyenv, benchmarked: uv installs 50x faster with a warm cache, conda still wins for CUDA, uv vs pyenv on version pinning. Plus a migration table.
- [uv Package Manager 2026: Install, Workflow, Migration Guide](https://www.heyuan110.com/posts/python/2026-04-10-uv-python-package-manager/): uv is a Rust-based Python package manager 10-100x faster than pip. How to install uv, set up projects and lock files, pin Python versions, and migrate from pip.
- [Conda Guide: How to Manage Multiple Python Versions and](https://www.heyuan110.com/posts/python/2020-01-11-python-conda/): Learn how to use Conda for Python environment management. This guide covers Conda vs pip, Anaconda vs Miniconda, creating and managing environments...

## TYPESCRIPT — 1


- [TypeScript Complete Guide: From Basics to Advanced Type](https://www.heyuan110.com/posts/typescript/2026-01-23-typescript-complete-guide/): A comprehensive TypeScript tutorial covering basic syntax, advanced types, generics, type gymnastics, utility types, and best practices with real-world code...
# ===== 中文文章 =====

## AI（中文） — 179


- [Windows 下好用的 Shell 推荐：别再把终端当 Shell 了](https://www.heyuan110.com/zh/posts/ai/2026-09-17-modern-windows-shell-terminal-guide/): 2026 年 Windows 命令行终极方案：彻底理清 Terminal 与 Shell 区别，日常主力用 PowerShell 7，数据处理用 Nushell，Linux 依赖走 WSL2。
- [Codex 接入第三方模型完美指南：深入 config.toml 与 CC Switch 协议转换](https://www.heyuan110.com/zh/posts/ai/2026-09-15-configuring-third-party-models-in-codex/): 解锁 DeepSeek、Claude 等三方模型在 OpenAI Codex 中的无限潜能。详尽拆解 Responses 协议差异、0.134.0+ 独立 profile 新规，以及 CC Switch 协议转换实操。
- [极简主义的胜利：Pi (pi.dev) Agent 双层架构与自研插件深度剖析](https://www.heyuan110.com/zh/posts/ai/2026-09-14-pi-coding-agent-deep-dive/): 深度拆解爆火极简 Coding Agent 工具 Pi (pi.dev) 的底层原理。详细剖析其双层循环机制、分支会话树、本地与全局插件隔离，以及如何用 AI 自主编写 TUI 与安全拦截插件。
- [深度驾驭 Anthropic Agent SDK 钩子：通过 pretoolus](https://www.heyuan110.com/zh/posts/ai/2026-09-14-mastering-anthropic-agent-hooks/): 深度剖析 Anthropic Agent SDK 的生命周期钩子。实战演练通过 pretooluse 和 posttooluse 拦截 Tool-calling 循环、防御提示词注入漏洞并进行 Tok...
- [斯坦福 CS146S 2026 秋季课程表：视频在哪、怎么免费跟课](https://www.heyuan110.com/zh/posts/ai/2026-09-11-cs146s-fall-2026-follow-along/): 斯坦福 CS146S 2026 秋季 9 月 22 日开课，12 月 3 日结束，官方没有课堂录像。本文给出完整课程表、能看的视频到底有哪些、和 2025 秋的差异，以及国内时区的逐周跟课方案。
- [ego lite 实测：把你登录好的浏览器交给 Claude Code，toke](https://www.heyuan110.com/zh/posts/ai/2026-09-10-ego-lite-browser-review/): ego lite v0.5 接 Claude Code 跑 3 个真实任务，对比 agent-browser 和 Chrome DevTools MCP：最好 16 秒 $0.12，最差 195 秒...
- [2026 秋季 5 门免费大学 AI Agent 课怎么跟：斯坦福、CMU、MI](https://www.heyuan110.com/zh/posts/ai/2026-09-09-free-ai-agent-courses-fall-2026/): 斯坦福 CS146S/CS329Z/CS329A、CMU 11-768、MIT 多模态课五门免费 AI Agent 课程横向对比：哪门有公开视频、哪门只有材料、国内怎么获取、先跟哪门、每周几小时。
- [CMU 11-768 AI Agents 课程拆解：OpenHands 作者教你](https://www.heyuan110.com/zh/posts/ai/2026-09-08-cmu-11-768-ai-agents-course/): CMU 11-768 AI Agents 2026 秋季课全拆解：Graham Neubig 与 Daniel Fried 的 28 节课表、三份作业（Harness/评测/RL 训练）、评分、免费视...
- [pi coding agent 实测：只给 4 个工具的极简 Agent，比 C](https://www.heyuan110.com/zh/posts/ai/2026-09-06-pi-coding-agent-review/): pi coding agent 与 Claude Code 跑同样 3 个任务实测：上下文 1,358 vs 31,012 token，4 个工具、不弹权限框、不带沙箱。附 DeepSeek / 通义...
- [斯坦福 CS329Z Engineering AI Agents 课程拆解：没有](https://www.heyuan110.com/zh/posts/ai/2026-09-04-stanford-cs329z-engineering-ai-agents/): 斯坦福 CS329Z（2026 秋）把 AI Agent 当工程学科教：拆解、数据、评测三大挑战。完整带日期课表、HW1/HW2 作业说明、评分表、和 CS146S 怎么二选一，以及没有公开录像时用...
- [lazygit 还是 gitui：终端 Git 工具选型（2026 实测）](https://www.heyuan110.com/zh/posts/ai/2026-09-02-gitui-vs-lazygit/): 8.2 万 commit 仓库实测：gitui 11 ms 出画面、全量历史只占 31 MB；lazygit 要 416 ms，却独占交互式 rebase、冲突解决、worktree 和自定义命令。附...
- [ubrowser 实测：号称最快最便宜的 Claude Code 浏览器自动化，](https://www.heyuan110.com/zh/posts/ai/2026-07-27-ubrowser-review/): 我把 lulzx/ubrowser 装起来跑了一遍：同一页面快照 761 token，Playwright MCP 要 7400。设计思想是对的，但这个项目本身有大问题，实测数据全在文里。
- [Claude Code 截图 MCP 配置:浏览器自动化调试省 100 倍 to](https://www.heyuan110.com/zh/posts/ai/2026-07-21-claude-code-screenshot-mcp-frontend-debugging/): 一条命令配好 Claude Code 截图 MCP,再告诉你 Chrome DevTools MCP 和 Playwright MCP 怎么选:实测 snapshot 10,220 token,定向脚...
- [OpenAI Codex App 2026：是什么、使用方法与避坑手册](https://www.heyuan110.com/zh/posts/ai/2026-07-12-openai-codex-app-guide/): OpenAI Codex 2026 是什么、怎么用？Codex App 已并入 ChatGPT 桌面端。手册讲清功能介绍、安装配置、使用方法、credit 计费与 CLI/Claude Code 怎么选。
- [Seedance 2.0 提示词最佳实践：从官方指南到实战](https://www.heyuan110.com/zh/posts/ai/2026-07-11-seedance-2-prompt-guide/): Seedance 2.0 提示词实战指南：五段式公式模板、@引用素材分工原则、六大翻车模式排查表、跨模型迁移对比与分镜到成片完整工作流，即梦和火山方舟均适用。
- [GPT-5.6 正式发布：三档价格、Codex 并入 ChatGPT Work](https://www.heyuan110.com/zh/posts/ai/2026-07-10-gpt-5-6-general-availability/): GPT-5.6 于 2026 年 7 月 9 日全量开放：Sol/Terra/Luna 三档定价对着 Claude 打、Codex 并入 ChatGPT Work 意味着什么、厂商跑分该打几折、国内怎...
- [Claude Fable 5 值不值：$10/$50 的决策账与用好它的方法](https://www.heyuan110.com/zh/posts/ai/2026-07-10-claude-fable-5-guide/): 2026 年 7 月 13 日起 Fable 5 退出所有 Claude 订阅，只能用 usage credits 按 $10/$50 计费。本文算清单次任务 $30-100 的真实成本账，给出该不该...
- [Seedream 5.0 Pro 上线：字节生图杀入 Gemini 腹地？API](https://www.heyuan110.com/zh/posts/ai/2026-07-09-seedream-5-pro/): Seedream 5.0 Pro 于 2026 年 7 月 8 日上线火山方舟：1K 单张 0.3 元、2K 0.6 元，按张计费。本文核检"超过 Gemini"的真实出处，给出开通、调用、计费全流程...
- [SpaceX 600 亿美元收购 Cursor:还能用吗?要不要迁移?](https://www.heyuan110.com/zh/posts/ai/2026-07-09-spacex-cursor-acquisition/): 已核实:SpaceX 于 2026 年 6 月 16 日以 600 亿美元全股票收购 Cursor 母公司 Anysphere,Q3 交割。本文分析 Claude 会不会被切、会不会涨价、国内开发者还...
- [Lovable vs v0 vs Bolt:2026 AI 应用生成器怎么选](https://www.heyuan110.com/zh/posts/ai/2026-07-08-lovable-vs-v0-vs-bolt/): Lovable、v0、Bolt 哪个好?不会写代码做全栈 MVP 选 Lovable,Vercel/Next.js 前端选 v0,要浏览器 IDE 和 Expo 移动端选 Bolt,会用终端直接上 C...
- [Claude 免费版额度 2026：能发多少条，何时升 Pro](https://www.heyuan110.com/zh/posts/ai/2026-07-08-claude-free-tier-limits/): Claude 免费版额度 2026 直接给答案：每 5 小时滚动窗口约 15-40 条消息，全天约 30-100 条，按 token 计费不按条数。免费版含 Sonnet 和联网搜索，但没有 Opus...
- [2026 最强 AI 编码模型对比：Fable 5、Sonnet 5 还是 GP](https://www.heyuan110.com/zh/posts/ai/2026-07-07-best-ai-coding-models-2026/): 2026 年写代码选哪个模型？直接答案：大多数人选 Sonnet 5——Terminal-Bench 反超旗舰 Opus 4.8，价格只有四成；Fable 5 留给一次顶一天工作量的任务。附选型决策树与价格速查表。
- [Loop Engineering 循环工程:给 AI Agent 造一个笼子](https://www.heyuan110.com/zh/posts/ai/2026-07-05-loop-engineering/): 循环工程(Loop Engineering)是 2026 年 AI Agent 工程栈的最外层学科。四大支柱——上下文保洁、刹车机制、毒舌评审、幂等工具——每根都配可直接抄的伪代码和真实翻车案例。核心...
- [MCP 落伍了？CLI + Skill 才是 Agent 工具链的未来](https://www.heyuan110.com/zh/posts/ai/2026-07-04-cli-skills-vs-mcp/): MCP 没死，但已从默认集成方式降级：GitHub MCP 光描述工具就吃约 5 万 token，一份 SKILL.md 只要 200。第一手生产实践讲透 CLI + Skill 为什么接管 Agen...
- [Agentic Loops 2026：让 AI Agent 自主循环干活](https://www.heyuan110.com/zh/posts/ai/2026-07-03-agentic-loops/): 讲清楚 agentic loop 和 Ralph loop 到底是什么，自主 agent 循环什么时候真能干活、什么时候烧钱失控，以及在 Claude Code / Cursor 上跑自主编码循环前必...
- [斯坦福 CS146S 自学手册 2026：逐讲笔记与实操路线](https://www.heyuan110.com/zh/posts/ai/2026-07-02-cs146s-study-guide/): 斯坦福 CS146S（The Modern Software Developer）自学手册：逐讲核心判断、必做练习与 2026 工具映射（Claude Code/Cursor），附两周速通与六周路线图。
- [OpenSpec vs Superpowers 实战工作流:spec 驱动开发怎](https://www.heyuan110.com/zh/posts/ai/2026-06-28-openspec-superpowers-workflow/): OpenSpec 和 Superpowers 一起装、零配置必打架:同一个功能一小时内长出两份互相矛盾的规格,apply 还写出零测试实现。本文给出我踩坑后沉淀的 CLAUDE.md 路由块和单用/组...
- [Windows 终端推荐 2026：5 款横评与选型决策树](https://www.heyuan110.com/zh/posts/ai/2026-06-22-best-windows-terminal-2026/): Windows 终端首选 Windows Terminal + PowerShell 7 + WSL2，九成人够用；跨平台换 WezTerm，极致速度选 Alacritty + Zellij，AI 工作流上 Warp。
- [上下文工程 2026：编码 Agent 该做减法而不是加法](https://www.heyuan110.com/zh/posts/ai/2026-06-16-context-engineering-2026/): 上下文工程 2026 实战：为什么删上下文比加上下文更有效——Sourcegraph 同任务实测 5K 精准检索打败 100K 摘要，precision@5 从 0.14 升到 0.48。附先卸载再摘...
- [Anthropic 9650 亿估值买的不是模型，是 Agent 基础设施：国内](https://www.heyuan110.com/zh/posts/ai/2026-06-12-anthropic-965b-managed-agents/): Anthropic 秘密递交 IPO，9650 亿美元估值对 470 亿年化收入只有 20 倍——投资人定价的是 Managed Agents 这套生产级 Agent 运行时，不是 Claude 模型...
- [Apple WWDC 2026 大转向：Siri 改用谷歌 Gemini，国行](https://www.heyuan110.com/zh/posts/ai/2026-06-09-apple-wwdc-2026-gemini-siri-pivot/): WWDC 2026 快评：Siri 改由谷歌 Gemini 驱动，AFM Cloud Pro 跑在谷歌云的 NVIDIA GPU 上，苹果自研 AI 叙事终结。谁赢谁输、国行 iPhone 用什么模型...
- [Harness Engineering 是炒作还是红利期？投资象限决策框架（60](https://www.heyuan110.com/zh/posts/ai/2026-05-08-harness-engineering-window-of-opportunity/): Harness Engineering 是炒作吗？这个 2026 年 2 月才诞生的词 7 周成为大厂共识：LangChain 不换模型只改 Harness，TerminalBench 2.0 从 5...
- [Hermes Agent v0.10 深度评测：113K star 的黑马是真自](https://www.heyuan110.com/zh/posts/ai/2026-04-24-hermes-agent-v010-deep-review/): Hermes Agent v0.10 发布 8 天、GitHub star 破 113K，成为 2026 年最快的开源 Agent 框架。本文实测三层记忆、118 技能、Tool Gateway，拆解...
- [Claude Managed Agents 发布 vs OpenClaw 被封](https://www.heyuan110.com/zh/posts/ai/2026-04-24-claude-managed-agents-vs-openclaw/): 4 月 4 日 Anthropic 封禁 OpenClaw，4 月 16 日官方推出 Managed Agents 公测。12 天组合拳不是巧合，是 Harness 层商业化围剿的明确信号。本文复盘时...
- [深度挖掘 wshobson/agents：33.9K Star 的 Claude](https://www.heyuan110.com/zh/posts/ai/2026-04-20-wshobson-agents-deep-dive/): 33.9K Star 的 wshobson/agents 全仓库盘点：79 个 Plugin、184 个 Agent、150 个 Skill 的逐个解析，10 个场景该装哪些、6 个别家没有的独家组件...
- [Harness 六层架构倒着建：80% 稳定性来自第 5、6 层](https://www.heyuan110.com/zh/posts/ai/2026-04-18-harness-six-layers-reverse-build/): Harness 六层架构并不平等：60 天生产实测显示第 5 层评估 +22%、第 6 层容错恢复 +18%，两层合计贡献约 80% 稳定性。倒着建（6→1）把迭代时间砍 3 倍、4 周稳定上线，附各...
- [Playwright CLI + Skill 三段式：把 AI 浏览器自动化做到](https://www.heyuan110.com/zh/posts/ai/2026-04-18-playwright-cli-skill-zero-token-automation/): 浏览器自动化的成本要分三段降——AI 探索（消耗 41% 上下文）、提炼成 Skill（5%）、固化成脚本（0 Token）。Playwright CLI vs MCP 4 倍 Token 差实测，3...
- [Claude Agent SDK 实战指南：3 行 Python 搭建生产级 A](https://www.heyuan110.com/zh/posts/ai/2026-04-17-claude-agent-sdk-guide/): Claude Agent SDK 用 3 行 Python 就能搭出能读写文件、执行命令的 AI Agent——LangChain 同样功能约需 80 行。详解 query() 与 ClaudeSDK...
- [2026 终端 AI 编程工具深度横评：Claude Code、Codex CL](https://www.heyuan110.com/zh/posts/ai/2026-04-14-terminal-ai-coding-tools-2026-comparison/): 2026 终端 AI 编程工具横评：Claude Code、Codex CLI、Gemini CLI、Aider 分属三条赛道——预算 $0 选 Gemini CLI（免费 1000 次/天）+ Ai...
- [Hermes Agent v0.9 评测 + 安装指南：Nous Researc](https://www.heyuan110.com/zh/posts/ai/2026-04-14-hermes-agent-guide/): Nous Research 开源 Hermes Agent v0.9.0 实战评测：4 月 13 日发布、GitHub 27000+ stars、$5 Hetzner VPS 起步、自改进 + 自写...
- [Mac 本地跑大模型实测：M3 Max 比 M4 Pro 强在哪](https://www.heyuan110.com/zh/posts/ai/2026-04-14-mac-apple-silicon-ai-workstation/): M3 Max Studio 128GB 跑 Llama 3.3 70B 实测 9.8 tok/s，M4 Pro 48GB 直接跑不动。决定速度的是内存带宽不是 GPU 核心数，附 48/64/128GB 选购建议。
- [OpenClaw 工具与 Skill 完全手册 2026：内置工具清单 + 自定](https://www.heyuan110.com/zh/posts/ai/2026-04-14-openclaw-tools-skills-reference/): OpenClaw 2026 内置工具完整参数速查，含 write tool 的 path 和 content 参数详解、ClawHub skill 加载机制、SKILL.md 格式、tavily-se...
- [Sub-Agent 架构设计：什么时候该拆子 Agent，Opus/Sonnet](https://www.heyuan110.com/zh/posts/ai/2026-04-13-harness-subagent-architecture/): Sub-Agent 架构的本质是上下文垃圾回收，不是并行加速器。实测经验：每次 spawn 至少 10000 token 输入才划算，Sonnet 做调度、Opus 做最终生成、Haiku 做扫描的路...
- [Lazygit：用了就回不去的终端 Git 神器](https://www.heyuan110.com/zh/posts/ai/2026-04-10-lazygit-guide/): Lazygit 把 interactive rebase、逐行 staging 这些痛苦的 Git 操作变成按键操作。76K stars、8 年迭代，这篇文章告诉你为什么它值得学、怎么用、什么时候不该用。
- [Claude HUD 是什么？3 分钟装好 Claude Code 状态栏插件（](https://www.heyuan110.com/zh/posts/ai/2026-04-10-claude-hud-guide/): Claude HUD 是 Claude Code 的开源状态栏插件，实时显示 context 用量、rate limit 消耗和 agent 状态，18000+ stars。本文给出 3 分钟安装命令...
- [Claude Code + OpenSpec + Superpowers：三件套](https://www.heyuan110.com/zh/posts/ai/2026-04-09-claude-code-openspec-superpowers/): 2026 年 Claude Code、OpenSpec、Superpowers 到底要不要一起上？实测后给出 trade-off 判断框架：三件套何时值回票价，何时只是徒增开销，以及每个工具真正独占的能力是什么。
- [一个人+AI Agent 月增 2000 客户：Paperclip 营销自动化实](https://www.heyuan110.com/zh/posts/ai/2026-04-04-ai-agent-marketing-automation/): AI Agent 营销自动化实战：Postiz 创始人用 Paperclip + Claude Code 搭起 4 个 AI 员工的虚拟营销部，一个人月增 2000 客户、MRR 做到 $45K，AP...
- [Claude Code Buddy 终端宠物：藏在 AI 编程工具里的电子宠物](https://www.heyuan110.com/zh/posts/ai/2026-04-04-claude-code-buddy-terminal-pet/): Claude Code Buddy 是内置在 Claude Code v2.1.89+ 的终端电子宠物：输入 /buddy 即可孵化，18 种物种、5 级稀有度，基于用户 ID 用 FNV-1a 哈希...
- [Claude 价格 2026：订阅与 API 六档套餐全对比](https://www.heyuan110.com/zh/posts/ai/2026-04-03-claude-pricing-complete-guide/): Claude Pro 每月 20 美元，Max 5x/20x 分别 100/200 美元，Team 每人 25 美元起，免费版 0 元只给 Sonnet。六档套餐的用量上限、API 计费与选型决策树一次看全。
- [5 款 AI 编程工具实测对比：为什么只选一个是错的](https://www.heyuan110.com/zh/posts/ai/2026-04-03-claude-code-vs-cursor-vs-copilot/): 用了 8 个月 Claude Code、Cursor、Copilot、Codex CLI、Gemini CLI 后的真实体验。每月 30 元的组合方案比 200 元的单工具好用。附决策框架和预算建议。
- [OpenClaw 多 Agent 配置实战：别让 LLM 做编排](https://www.heyuan110.com/zh/posts/ai/2026-04-02-openclaw-multi-agent-setup-guide/): 手把手配置 OpenClaw 多 Agent 系统。8 级 binding 优先级详解、Lobster 确定性流水线 vs sessions_send 对比、飞书集成、成本控制策略，附可复制的 3 Agent 起步配置。
- [MCP、Skills、Hooks 到底有什么区别？Claude Code 三大扩](https://www.heyuan110.com/zh/posts/ai/2026-04-02-mcp-vs-skills-claude-code/): MCP、Skills、Hooks 的区别一文讲透：MCP 连外部服务提供能力，Skill 是教 AI 做事的知识包，Hooks 强制执行硬规则。一个 15 工具的 MCP 服务器常驻占约 1500-3...
- [Cursor Composer 2 深度评测：Kimi K2.5 风波背后的真相](https://www.heyuan110.com/zh/posts/ai/2026-04-01-cursor-composer-2-review/): Cursor Composer 2 深度评测：2026 年 3 月 19 日发布，三天后被发现底座是月之暗面 Kimi K2.5，75% 算力做 RL 的说法站得住吗？Terminal-Bench 6...
- [Claude Code 开源风波：源码泄露到社区重写的全过程](https://www.heyuan110.com/zh/posts/ai/2026-04-01-claude-code-open-source-agent/): 深度技术分析 Claude Code 开源版 — 基于 Claude Code 泄露架构的 Python + Rust 开源重写。涵盖源码泄露事件始末、架构对比、法律风险评估，以及是否值得从 Clau...
- [CLAUDE.md 最佳实践：我把 90 行砍到 50 行，Agent 表现反而](https://www.heyuan110.com/zh/posts/ai/2026-03-31-harness-claudemd-guide/): 一个真刀真枪的 CLAUDE.md 写法指南：我翻车的 90 行版本，ETH Zurich 研究证实的 60 行上限，AI 自动生成降 20%。含分层架构、3 个误区、90 秒审查清单和可拷贝的 50 行模板。
- [Harness Engineering 实战：模型是 AI Agent 里最不重](https://www.heyuan110.com/zh/posts/ai/2026-03-30-harness-engineering-guide/): 60 天跑 Claude Code 博客管线，换模型只升 5%，重写 Harness 端到端降本 60%。讲透 Agent = 模型 + Harness 公式、引导器+传感器架构、Sub-Agent...
- [Seedance 2.0 技术深度拆解：字节跳动如何做出排名第一的AI视频模型](https://www.heyuan110.com/zh/posts/ai/2026-03-29-seedance-2-bytedance-ai-video/): Seedance 2.0 深度拆解：字节跳动 2026 年 2 月发布的 AI 视频模型以 Elo 1269 登顶 Artificial Analysis 盲测榜，击败 Veo 3 和 Sora 2...
- [Lark CLI 完全指南：用命令行和 AI Agent 操控飞书](https://www.heyuan110.com/zh/posts/ai/2026-03-29-lark-cli-guide/): Lark CLI 是飞书开放平台官方命令行工具，覆盖日历、消息、文档等 11 大业务域，提供 200+ 命令和 19 个 AI Agent Skills。本文手把手教你从安装配置到实战操作，让 AI...
- [内网穿透实战：SSH 隧道、frp、Cloudflare Tunnel 三种方案](https://www.heyuan110.com/zh/posts/ai/2026-03-28-cloudflare-tunnel-guide/): 内网穿透怎么选？SSH 反向隧道、frp、Cloudflare Tunnel 三种方案深度对比：前两种需要一台有公网 IP 的 VPS，Cloudflare Tunnel 免费但要托管域名。附一条 s...
- [QMD：本地语义搜索引擎，帮 AI Agent 省下 90% Token 开销](https://www.heyuan110.com/zh/posts/ai/2026-03-25-qmd-local-search-ai-agent-memory/): QMD 是 Shopify CEO Tobi Lütke 开源的本地语义搜索引擎：BM25 + 向量检索 + LLM 重排序三层架构，全部本地运行、模型总共不到 2GB，让 AI Agent 按需精准...
- [Chrome DevTools MCP 2026 配置教程：彻底解决「打开新窗口](https://www.heyuan110.com/zh/posts/ai/2026-03-17-chrome-devtools-mcp-guide/): 手把手教你让 AI 连接你已登录的 Chrome 浏览器：3 步修复「每次开新窗口」，配置 9222 远程调试端口（macOS/Windows 全覆盖），Chrome 146+ 启用 autoConn...
- [GitHub Agent HQ + VS Code 完整指南 2026：怎么用？](https://www.heyuan110.com/zh/posts/ai/2026-03-13-github-agent-hq-vscode-guide/): GitHub Agent HQ 实战教程：VS Code v1.109 配置、Claude/Codex/Copilot 三 Agent 并行对比、本地/云端/后台三种模式。含 Copilot Pro+...
- [2026年AI代码安全工具横评：Codex Security、Claude Co](https://www.heyuan110.com/zh/posts/ai/2026-03-13-ai-code-security-tools-compared/): AI代码安全工具横评 2026：Codex Security（3 月发布，扫描 120 万次提交发现 792 个严重漏洞、14 个 CVE）、Claude Code Security（2 月发布，基于...
- [GPT-5.4 vs Claude Opus 4.6 终极对比 2026：编程/](https://www.heyuan110.com/zh/posts/ai/2026-03-13-gpt-5-4-vs-claude-opus-comparison/): 一文看懂 2026 AI 模型选型。GPT-5 和 Claude 到底哪个强？实测 SWE-bench Pro（57.7% vs 45.9%）、API 价格（便宜一半）、100 万上下文、Agent...
- [OpenCode 深度评测：这款开源AI编程代理能替代Claude Code吗？](https://www.heyuan110.com/zh/posts/ai/2026-03-13-opencode-ai-coding-agent-review/): OpenCode 深度评测：两周实测这个 GitHub 12 万星的 MIT 开源 AI 编程代理——支持 75+ 模型提供商、LSP 自修正是杀手锏，但客户端-服务器架构曾曝出 CVSS 约 10...
- [AI 编程方法论对比 2026：Vibe Coding vs SDD vs BM](https://www.heyuan110.com/zh/posts/ai/2026-03-11-ai-development-methodologies-compared/): 6 大 AI 开发方法论终极对比：Vibe Coding、SDD 规范驱动、BMAD、Ralph Loop、Peter 工作流和务实混合流。附 Martin Fowler 评价、YC 25% 公司数据...
- [TypeScript vs Python：AI 时代该选哪门语言？2026 全面](https://www.heyuan110.com/zh/posts/ai/2026-03-10-typescript-vs-python-ai-era/): TypeScript vs Python 2026 全面对比：TypeScript 以 66% 增速登顶 GitHub（月活 263 万），Python 增长 48% 但仍占近 50% 的新建 AI...
- [高频提交：如何做到日均 100+ Commit 而不翻车](https://www.heyuan110.com/zh/posts/ai/2026-03-10-high-frequency-commits/): 日均 100+ 次 Git 提交怎么不翻车？AI 编程让团队日产 50-150 次提交成为常态，答案不是放慢速度而是建护栏：原子提交、约定式提交、5-15 分钟内返回结果的分层测试 CI、功能开关与渐...
- [AI 开发环境搭建：2026 年工具、配置与 Dotfiles 完全指南](https://www.heyuan110.com/zh/posts/ai/2026-03-10-ai-dev-environment-setup/): AI 开发环境搭建 2026 完全指南：Claude Code 负责规划推理 + Cursor 负责日常编码的组合月均约 40 美元即可覆盖几乎所有场景。涵盖 CLAUDE.md、MCP 服务器、tm...
- [上下文工程：2026年最被低估的AI开发技能](https://www.heyuan110.com/zh/posts/ai/2026-03-10-context-engineering-guide/): 上下文工程是 2026 年最被低估的 AI 开发技能：研究显示上下文超过约 3000 token 推理质量就开始下降，指令最佳长度是 150-300 个单词。深入拆解信息选择等五大维度、污染/干扰/混...
- [从原型到生产：AI 应用上线完整检查清单](https://www.heyuan110.com/zh/posts/ai/2026-03-09-prototype-to-production/): AI 应用从原型到生产的完整检查清单：构建原型只占总工作量的 10%，剩下 90% 才是产品与 Demo 的分界线。逐一拆解测试覆盖、安全加固、可扩展性、可观测性、自动化运维、持续演进六道关卡，附各阶...
- [MCP 安全危机：60 天内曝出 30 个 CVE，到底怎么了？](https://www.heyuan110.com/zh/posts/ai/2026-03-10-mcp-security-2026/): MCP 安全危机实录：2026 年 1-2 月 60 天内曝出 30+ 个 CVE，受影响下载 43.7 万次，被扫描的 2614 个 MCP 实现中 82% 存在路径穿越风险、43% 的漏洞是 Sh...
- [TypeScript 暴涨 66% 的真相：AI 编程工具正在重塑语言选择](https://www.heyuan110.com/zh/posts/ai/2026-03-10-typescript-ai-tools/): TypeScript 以 66.6% 年增长率登顶 GitHub 第一语言，月活贡献者达 263 万，一年内超越 Python 和 JavaScript。背后推手是 AI 编程工具：研究发现 94%...
- [Cursor 安装教程 2026：官方下载、配置到 Agent 模式实战](https://www.heyuan110.com/zh/posts/ai/2026-03-08-cursor-setup-guide/): Cursor 安装教程 2026：从 cursor.com 官方下载、安装到 Agent 模式——Shift+Tab 规划模式、.cursor/rules/ 规则、Git Worktrees 多 Agent 并行。
- [AI 编程工具 2026 排名：7 款主流工具实测对比](https://www.heyuan110.com/zh/posts/ai/2026-03-10-ai-coding-agents-comparison-2026/): 实测对比 Claude Code、Cursor、Google Antigravity、GitHub Copilot、Kiro、Codex CLI、Windsurf 七款工具的价格、功能与基准测试表现。
- [Kiro 深度评测：亚马逊这款「规格驱动」的 AI 编程代理到底行不行？](https://www.heyuan110.com/zh/posts/ai/2026-03-10-kiro-review/): Kiro 是亚马逊的规格驱动 AI 编程 IDE：先生成规格文档再写代码，Pro 版 19 美元/月，基于 Code OSS 兼容 VS Code 插件。深度评测 Agent Hooks、Steeri...
- [用 Python 从零构建 AI 编程 Agent（完整实战教程）](https://www.heyuan110.com/zh/posts/ai/2026-03-07-build-ai-agent-python/): 用 Python 从零构建 AI 编程 Agent：250 行代码实现 Agentic Loop + Function Calling，带读写文件、执行命令、搜索代码等 6 个工具。从 20 行的基础...
- [Codex CLI 深度指南：安装配置、安全模型与 20+ 高手技巧](https://www.heyuan110.com/zh/posts/ai/2026-03-10-codex-cli-deep-dive/): Codex CLI 深度指南：默认模型 gpt-5.3-codex，需要 ChatGPT Plus/Pro（$20-200/月）或 API Key。详解 Auto/Read Only/Full Acc...
- [Google Antigravity 深度测评：这款免费 Agent-First](https://www.heyuan110.com/zh/posts/ai/2026-03-10-google-antigravity-review/): Google Antigravity IDE 深度体验报告——基于 Gemini 3 的免费 Agent-First 开发平台。涵盖安装配置、核心功能、Manager View 以及与 Cursor...
- [OpenAI Symphony：从 Issue 到 PR 全自动，开发者不用动手](https://www.heyuan110.com/zh/posts/ai/2026-03-05-openai-symphony-autonomous-coding/): OpenAI Symphony 监控你的工单系统，自动派 Codex Agent 写代码、跑测试、提 PR。架构拆解、WORKFLOW.md 配置、真实局限性分析。
- [OpenClaw vs AutoGPT vs CrewAI：2026年最佳个人](https://www.heyuan110.com/zh/posts/ai/2026-03-05-openclaw-vs-ai-agents/): 深度对比 OpenClaw AI agent framework 与 AutoGPT、CrewAI、LangGraph、AutoGen、Devin 六大方案（2026），从架构、成本、多智能体支持到消...
- [OpenClaw ClawHub Tavily Search Skill 集成完](https://www.heyuan110.com/zh/posts/ai/2026-03-05-openclaw-tavily-integration/): 通过 ClawHub 安装 tavily-search skill，为 OpenClaw agent 接入网络搜索仅需三步。含 clawdhub install 命令、API Key 配置、按 Age...
- [OpenClaw 踩坑指南：15 个自动化常见错误及解决方案](https://www.heyuan110.com/zh/posts/ai/2026-03-05-openclaw-automation-pitfalls/): 全面梳理 OpenClaw 自动化中最容易踩的 15 个坑，涵盖配置、Agent 行为、成本控制和安全四大类，附带真实场景和修复方案。
- [Claude Code 斜杠命令、快捷键与 CLI 完全参考手册](https://www.heyuan110.com/zh/posts/ai/2026-03-05-claude-code-slash-commands/): 全面整理 Claude Code 的每个斜杠命令、键盘快捷键、CLI 参数和环境变量，以及如何用 Skills 创建自定义命令。
- [用 Python 构建 MCP 服务器：完整分步教程](https://www.heyuan110.com/zh/posts/ai/2026-03-05-build-mcp-server-python/): 用 Python 构建 MCP 服务器分步教程：基于 FastMCP 装饰器 API，不到 50 行代码就能跑起一个服务器，要求 Python 3.10+。涵盖工具/资源/提示词三种能力、MCP In...
- [OpenClaw 安装配置完全指南：从零搭建你的私人 AI Agent](https://www.heyuan110.com/zh/posts/ai/2026-03-05-openclaw-setup-guide/): OpenClaw 完整安装教程，涵盖一键安装、Telegram/WhatsApp 消息平台对接、AI 模型配置、Skills 插件管理、多 Agent 协作入门以及安全加固最佳实践。
- [OpenClaw 多 Agent 搭建指南：让 AI 团队协作起来](https://www.heyuan110.com/zh/posts/ai/2026-03-05-openclaw-multi-agent-setup/): OpenClaw 多 Agent 搭建完整教程。涵盖 Agent 团队配置、消息路由绑定、Tavily 搜索集成、Agent 间通信，以及 4 种协作模式的实战讲解。
- [Claude Code 和 Copilot 哪个好？用了三个月的真实对比结论](https://www.heyuan110.com/zh/posts/ai/2026-03-05-claude-code-vs-copilot/): 结论先行：大型重构用 Claude Code，日常补全用 Copilot，两者可同时跑。三个月双持实测：补全速度、Agent 能力、价格、真实代码示例全维度对比，帮你把钱花对地方。
- [什么是 Claude Code Hooks？17 个生命周期事件入门详解](https://www.heyuan110.com/zh/posts/ai/2026-03-05-claude-code-hooks-guide/): 零基础理解 Claude Code Hooks：它是什么、怎么工作，17 个生命周期事件逐一讲解，Hook 类型、matcher 规则与退出码全解析，新手入门首选参考。
- [CLAUDE.md 最佳实践：写出真正有效的配置文件](https://www.heyuan110.com/zh/posts/ai/2026-03-05-claude-code-claudemd-best-practices/): 如何为 Claude Code 编写高效的 CLAUDE.md 文件。实用模式、7 种项目类型示例，以及 CLAUDE.md vs .cursorrules vs AGENTS.md 对比。
- [2026 年最值得装的 18 个 MCP 服务器：Claude Code 实测精](https://www.heyuan110.com/zh/posts/ai/2026-03-05-best-mcp-servers-claude-code/): 2026 年最值得装的 MCP 服务器实测精选：从 10000+ 个里挑出 18 个真正实用的，覆盖 PostgreSQL 数据库直连、Playwright 浏览器自动化、GitHub 操作与 Con...
- [Claude Code 价格 2026：免费/Pro $20/Max $200](https://www.heyuan110.com/zh/posts/ai/2026-02-25-claude-code-pricing/): Claude Code 价格 2026 真实对比：免费版不含 Claude Code，Pro $20/月每 5 小时约 45 条消息，Max 5x $100 约 225 条，Max 20x $200...
- [OpenClaw 2026.3.1 新功能详解：WebSocket 流式传输、A](https://www.heyuan110.com/zh/posts/ai/2026-03-03-openclaw-2026-3-1-new-features/): OpenClaw 2026.3.1 新版本完整指南，涵盖 WebSocket 传输、Agent 路由 CLI、外部密钥管理、Kubernetes 健康检查端点以及 Claude 4.6 自适应思考等全部新功能。
- [Tmux 完全指南：从入门到 AI 多智能体并行开发工作流](https://www.heyuan110.com/zh/posts/ai/2026-03-03-tmux-guide-ai-development/): 从零开始掌握 tmux，学会会话管理、窗格分屏、自定义配置，以及如何使用 tmux 运行多个并行 Claude Code 智能体，实现 10 倍开发效率提升。
- [Vibe Coding 详解：它是什么，怎样才能用好？](https://www.heyuan110.com/zh/posts/ai/2026-02-28-vibe-coding-explained/): 什么是 Vibe Coding？由 Andrej Karpathy 提出的 AI 开发方式。了解其工作流程、最佳工具、安全风险以及 2026 年的最佳实践。
- [MCP 协议详解：AI 工具的通用连接标准](https://www.heyuan110.com/zh/posts/ai/2026-02-28-mcp-protocol-explained/): 深入解析 Model Context Protocol (MCP) 的架构、核心概念、生态系统，以及 MCP 与 Function Calling 的区别。2026 年 AI 开发必备知识。
- [Claude vs ChatGPT vs Gemini：2026年最佳编程LLM](https://www.heyuan110.com/zh/posts/ai/2026-03-02-claude-vs-chatgpt-vs-gemini/): 深度对比 Claude Opus 4.6、GPT-5.2 和 Gemini 2.5 Pro 的编程能力。真实基准测试、定价、上下文窗口和使用场景推荐，帮你选出最适合项目的 LLM。
- [GitHub Copilot vs Claude Code vs Cursor](https://www.heyuan110.com/zh/posts/ai/2026-02-28-copilot-vs-claude-vs-cursor/): 深度对比 GitHub Copilot、Claude Code 和 Cursor 三大 AI 编程工具，从定价、Agent 能力、代码补全到 IDE 支持全面分析，帮你找到 2026 年最适合的 AI 编程助手。
- [用 TypeScript 构建 MCP Server：从零到发布完整教程](https://www.heyuan110.com/zh/posts/ai/2026-03-02-building-mcp-servers-typescript/): 手把手教你用 TypeScript 构建 MCP Server。学习如何使用 MCP SDK v2 创建工具、资源和提示模板，用 Claude Code 测试调试，最终发布到 npm。
- [Claude Code vs Cursor 2026：哪个 AI 编程工具更强？](https://www.heyuan110.com/zh/posts/ai/2026-02-28-claude-code-vs-cursor/): Claude Code 与 Cursor 2026 年深度对比：从 Agent 能力、代码质量、开发体验、定价到最佳场景，五个维度全面测评，帮你选出最适合的 AI 编程工具。
- [Claude Code 是什么？2026 完全指南：安装到工作流](https://www.heyuan110.com/zh/posts/ai/2026-02-28-claude-code-complete-guide/): Claude Code 是 Anthropic 的终端 AI 编程智能体，能读懂代码库、跨文件改代码、跑测试并自动迭代。安装配置、CLAUDE.md、MCP、Hooks、Skills、Worktree 与定价选型全覆盖。
- [RAG 管道搭建指南：向量数据库 + LLM 集成实战](https://www.heyuan110.com/zh/posts/ai/2026-03-01-rag-pipeline-setup/): 完整的 RAG 管道搭建教程，涵盖向量数据库选型、Embedding 策略、文本分块方法，以及基于 OpenAI、Chroma 和 Qdrant 的 Python 代码实战。
- [Claude 速率限制 2026：免费版/Pro/Max 每5小时消息上限详解](https://www.heyuan110.com/zh/posts/ai/2026-02-28-claude-rate-limits/): Claude 免费版限制与 Pro 限制 2026 实测：免费版每 5 小时约 2-5 条，Pro 约 10-45 条，Max 约 50-800 条，另有 7 天周限额，附滚动窗口机制与限速应对策略。
- [Claude Code MCP 配置指南：让 AI 连接任何外部服务](https://www.heyuan110.com/zh/posts/ai/2026-02-28-claude-code-mcp-setup/): 完整的 MCP Server 配置教程——安装社区服务器、用 TypeScript 从零构建自定义服务器，以及高效调试技巧。
- [CLAUDE.md 完全指南：让 AI 每次都精准理解你的项目](https://www.heyuan110.com/zh/posts/ai/2026-02-28-claude-code-claudemd-guide/): 深入掌握 CLAUDE.md 配置方法，让 Claude Code 每次会话都自动获取完整项目上下文。涵盖三层配置体系、实用模板和真实项目案例。
- [Claude Code Worktree：多任务并行开发完全指南](https://www.heyuan110.com/zh/posts/ai/2026-02-28-claude-code-worktree-guide/): 使用 Git worktree 模式同时运行多个 Claude Code 会话，实现并行开发、自动清理和团队协作工作流，大幅提升 AI 编程效率。
- [AI Agent 安全指南：2026年自动化工作流防护全攻略](https://www.heyuan110.com/zh/posts/ai/2026-02-27-ai-agent-security/): 深入解析 AI Agent 面临的安全威胁：提示注入、工具投毒、MCP 漏洞，涵盖 OWASP Agentic Top 10 框架、真实 CVE 案例、防御策略与安全工具推荐。
- [Claude Code --teammate-mode 详解：让多个 Agent](https://www.heyuan110.com/zh/posts/ai/2026-02-28-claude-code-teams-guide/): Claude Code Agent Teams 实战：设置 CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1 即可启用 2026 年 2 月随 Opus 4.6 发布的多...
- [Claude Code Skills 完全指南：用 SKILL.md 打造自定义](https://www.heyuan110.com/zh/posts/ai/2026-02-28-claude-code-skills-guide/): 手把手教你创建 Claude Code Skills，包含 SKILL.md 编写方法、3 个即用模板、社区热门 Skills 推荐，以及 Skills 与 CLAUDE.md、Hooks 的协同使用技巧。
- [Claude Code Hooks 详解：PreToolUse、PostTool](https://www.heyuan110.com/zh/posts/ai/2026-02-28-claude-code-hooks-guide/): PreToolUse、PostToolUse 在 .claude/settings.json 里怎么配：事件触发时机、matcher 规则、退出码含义全部讲透，附 12 个生产级配置和常见踩坑。
- [多智能体编排：4 种真正有效的架构模式](https://www.heyuan110.com/zh/posts/ai/2026-02-26-multi-agent-orchestration/): 深入解析 4 种经过生产验证的多智能体编排模式 — 流水线、路由器、编排者-工作者、评估器-优化器 — 结合 Claude Code、Cursor 和 Antigravity 的实战案例。
- [Claude Code 10 大常见坑 2026：新手必看的避坑指南（省 50%](https://www.heyuan110.com/zh/posts/ai/2026-02-25-claude-code-mistakes/): Claude Code 新手最爱踩的 10 个坑：没写 CLAUDE.md、默认用 Opus 烧钱、忽略 Hooks、权限弹窗烦人、超长对话不压缩……每条附具体解决方案，帮你省一半 Token、输出质...
- [Claude Code 安装配置完全指南 2026：10 分钟从零上手](https://www.heyuan110.com/zh/posts/ai/2026-02-25-claude-code-setup-guide/): Claude Code 2026 最新安装教程：macOS/Windows 终端一条命令搞定、API Key 与 OAuth 登录、VS Code/JetBrains 集成、CLAUDE.md 配置...
- [从零开始用 Python 构建你自己的 Claude Code（仅 250 行）](https://www.heyuan110.com/zh/posts/ai/2026-02-24-build-magic-code/): 一篇实战教程，通过用 Python 从零重建 Claude Code 的核心架构——Agentic Loop、Tool Use 和流式输出，彻底揭开 AI 编程助手的神秘面纱。从 20 行代码到完整的...
- [斯坦福 CS146S 精读（五）：从原型到生产——AI 应用完整生命周期](https://www.heyuan110.com/zh/posts/ai/2026-02-24-prototype-to-production/): 深度解读斯坦福 CS146S 第八九周课程：一句话做 App 只是起点，如何将 AI 快速原型纳入测试、安全、运维的完整工程体系，从 Demo 到 Production 的完整路径。
- [斯坦福 CS146S 精读（四）：Secure Vibe Coding——AI](https://www.heyuan110.com/zh/posts/ai/2026-02-24-secure-vibe-coding/): 深度解读斯坦福 CS146S 第六七周课程：Prompt Injection 导致远程代码执行的真实案例、OWASP Top 10 在 Agent 时代的新威胁、AI 代码审查方法论，以及如何建立安全...
- [斯坦福 CS146S 精读（三）：Agent Manager——人机协作的最佳实](https://www.heyuan110.com/zh/posts/ai/2026-02-24-agent-manager-patterns/): 深度解读斯坦福 CS146S 第四周课程：从 Claude Code 创始人 Boris Cherney 的演讲到 Anthropic 官方最佳实践，系统学习 Coding Agent 的自治度管控和人机协作模式设计。
- [斯坦福 CS146S 精读（二）：上下文工程——AI 编程最被低估的核心能力](https://www.heyuan110.com/zh/posts/ai/2026-02-24-context-engineering-deep-dive/): 深度解读斯坦福 CS146S 第三周课程：从 Prompt Engineering 到 Context Engineering 的范式跃迁，Spec 为何成为新的源代码，长上下文的四种失败模式与应对策略。
- [斯坦福 CS146S 全解析 2026：AI 编程公开课中文笔记（免费自学路线）](https://www.heyuan110.com/zh/posts/ai/2026-02-24-stanford-cs146s-overview/): 斯坦福顶级 AI 编程课 CS146S The Modern Software Developer 完整拆解：10 周大纲、Boris Cherney/Karpathy 级嘉宾阵容、免费 PPT 和作...
- [OpenClaw 多 Agent 配置教程 2026：从单 Agent 到团队协](https://www.heyuan110.com/zh/posts/ai/2026-02-23-openclaw-multi-agent-guide/): 手把手基于 OpenClaw AI agent framework（2026）搭建多 Agent 系统：路由绑定、Agent 间通信、四大协作模式（管道/广播/投票/分层）详解，含生产环境最佳实践和踩坑经验。
- [2026 Agentic Coding 趋势报告：Claude Code 年入](https://www.heyuan110.com/zh/posts/ai/2026-02-23-agentic-coding-trends-2026/): 深度解读 Anthropic 2026 Agentic Coding 趋势报告：从 Claude Code 6个月营收10亿到 Anthropic 年化140亿，拆解 AI 编程 8 大趋势、市场格局与开发者采用数据。
- [MCP 安全实战指南：AI Agent 时代的攻防博弈与防护策略](https://www.heyuan110.com/zh/posts/ai/2026-02-23-mcp-security-guide/): MCP 安全全解析：从 OWASP Top 10、CVE 漏洞案例到 mcp-scan 工具实战，一文掌握 AI Agent 时代的 MCP 安全攻防策略与最佳实践。
- [用 Claude Code 从零开发 MCP Server：TypeScript](https://www.heyuan110.com/zh/posts/ai/2026-02-22-claude-code-mcp-server-tutorial/): 手把手教你用 Claude Code 开发自定义 MCP Server，从项目初始化到工具定义、调试、发布，附完整代码和避坑指南。适合想给 AI 扩展能力的开发者。
- [Claude Code Security 深度解析：AI 驱动的代码安全扫描如何](https://www.heyuan110.com/zh/posts/ai/2026-02-22-claude-code-security/): 全面解析 Anthropic 最新发布的 Claude Code Security：AI 漏洞扫描原理、与传统 SAST 对比、实测发现 500+ 生产级漏洞、获取方式与行业影响分析。
- [Claude Code 2026年2月更新：Worktree 并行开发、后台任务](https://www.heyuan110.com/zh/posts/ai/2026-02-22-claude-code-february-updates/): Claude Code 2月迎来重磅更新：Git Worktree 隔离开发、后台 Agent 管理、Simple Mode 文件编辑等新功能。本文逐个解析用法和实战场景，附命令速查。
- [Vibe Coding 是什么？氛围编程完全指南与工具对比](https://www.heyuan110.com/zh/posts/ai/2026-02-22-vibe-coding-guide/): Vibe Coding 是 Karpathy 于 2025 年 2 月提出的编程方式：用自然语言描述需求，AI 生成实现，人只管方向和质量。含 Claude Code/Cursor/Trae 横评与完整实战流程。
- [Claude Code 多 Agent 协作教程 2026：Agent Team](https://www.heyuan110.com/zh/posts/ai/2026-02-22-claude-code-agent-teams/): 手把手配置 Claude Code Agent Teams：多 Agent 并行开发原理、Team Lead + Teammate 架构、Opus+Sonnet 混合编排省钱、命令速查表，将小时级工作压缩到分钟级。
- [AI Agent 记忆系统全解析：RAG 与上下文工程的对比与选择](https://www.heyuan110.com/zh/posts/ai/2026-02-21-ai-agent-memory-systems/): RAG 和上下文工程是 AI Agent 记忆系统的两大主流方案。本文深入对比两者的原理、优劣和适用场景，并提供实战落地方案，帮你为 LLM Agent 构建高效的持久化记忆。
- [Claude Code Worktree 实战：一个仓库同时跑多个 AI 任务的](https://www.heyuan110.com/zh/posts/ai/2026-02-20-claude-code-worktree/): 详解 Claude Code --worktree (-w) 模式，从 Git Worktree 基础到并行开发实战，涵盖自动创建、隔离开发、清理机制、最佳实践与常见问题。帮你在一个仓库里同时跑多个...
- [MCP 协议全面解析：AI 连接万物的通用标准](https://www.heyuan110.com/zh/posts/ai/2026-02-20-mcp-protocol-guide/): MCP 协议（Model Context Protocol）被称为 AI 领域的 USB-C：2024 年 11 月由 Anthropic 发布，现已捐赠 Linux Foundation，SDK 月...
- [Xcode 26.3 Agentic Coding 深度解析：Apple 如何用](https://www.heyuan110.com/zh/posts/ai/2026-02-20-xcode-agentic-coding/): Apple 发布 Xcode 26.3，首次集成 Anthropic Claude Agent 和 OpenAI Codex，支持 MCP 协议。本文深度解析其工作原理、实际体验和对 iOS 开发者的影响。
- [Figma Code to Canvas：Anthropic 联手打造代码转设计](https://www.heyuan110.com/zh/posts/ai/2026-02-19-figma-code-to-canvas/): Figma 与 Anthropic 合作推出 Code to Canvas 功能，将 Claude Code 生成的代码直接转化为可编辑的 Figma 设计稿。本文深度解析工作原理、使用方法与行业影响。
- [Claude Code 和 Codex CLI 哪个好？2026 八维度实测](https://www.heyuan110.com/zh/posts/ai/2026-02-19-claude-code-vs-codex/): Claude Code（Opus 4.6）对比 Codex CLI（GPT-5.3）八个维度：编码质量、百万 token 上下文、Agent 协作、安全沙箱与定价，两者入门价都是 $20/月。
- [Claude Code vs Cursor vs Windsurf（2026）实](https://www.heyuan110.com/zh/posts/ai/2026-02-18-claude-code-vs-cursor-vs-windsurf-2026/): 基于真实开发任务对 Claude Code、Cursor、Windsurf 做 2026 年实测对比，重点拆解速度、成本、可控性、学习成本与适用人群，并给出按场景可直接套用的选型建议。
- [Claude Code Hooks 配置大全：12 个可直接复制的自动化片段](https://www.heyuan110.com/zh/posts/ai/2026-02-18-claude-code-hooks-guide/): 12 个拿来即用的 Claude Code Hook 配置：保存自动格式化、.env 敏感文件保护、危险命令拦截、Git 自动暂存、Slack 通知，附完整团队组合配置方案。
- [MoltBot 是什么？改名 OpenClaw 全解析：原理、功能与上手教程](https://www.heyuan110.com/zh/posts/ai/2026-02-18-what-is-moltbot/): MoltBot 是什么？为什么改名 OpenClaw？本文详解 MoltBot 原理、核心功能、三次改名历史、与 Claude Code 等 AI Agent 对比，附快速上手教程。
- [Claude Code + Draw Things：Mac 本地 AI 自动配图](https://www.heyuan110.com/zh/posts/ai/2026-02-16-claude-code-draw-things-workflow/): 深度教程：用 Claude Code 通过 MCP 协议调用 Draw Things 实现本地 AI 自动配图。涵盖配置、4 大工具详解、自动化工作流实战，告别云端付费，Mac 一台搞定。
- [OpenClaw 的 30 天狂飙：180K Star、40+ 漏洞、创始人加入](https://www.heyuan110.com/zh/posts/ai/2026-02-16-openclaw-openai-analysis/): 深度复盘 OpenClaw AI agent framework 从诞生到创始人加入 OpenAI 的 30 天旅程（2026），解析 Skills 生态、Moltbook 社交网络、安全隐忧，以及这...
- [Draw Things 教程：Mac 免费 AI 生图与模型推荐](https://www.heyuan110.com/zh/posts/ai/2026-02-15-draw-things-ultimate-guide/): 免费 macOS 原生生图 App，靠 Metal FlashAttention 比 ComfyUI 快 20-40%。16GB Mac Mini M4 能跑 Flux、训 LoRA，附模型推荐与 MCP 集成。
- [Mac Mini 跑 ComfyUI 实测：24GB 出 Flux 图 50 秒](https://www.heyuan110.com/zh/posts/ai/2026-02-15-mac-mini-local-image-generation/): Mac Mini M4 Pro 24GB 跑 Flux 1024×1024 只要 50 秒。实测 ComfyUI、Draw Things、DiffusionBee，附 16/24/48GB 内存选购建议。
- [拆解 OpenClaw 自动化架构：从消息到执行的完整链路](https://www.heyuan110.com/zh/posts/ai/2026-02-14-openclaw-architecture-deep-dive/): 从架构视角拆解 OpenClaw AI agent framework（2026）的自动化原理：消息如何经由 Gateway 路由到 Agent，如何用 Skills 调工具、跨设备 Nodes 执行...
- [OpenClaw 自动化别踩坑：装 3 个 Skill 不等于真的好用](https://www.heyuan110.com/zh/posts/ai/2026-02-14-openclaw-automation-pitfalls/): 很多人以为给 OpenClaw 装上 tavily-search、find-skills、proactive-agent 就能自动化起飞。真正决定可用性的，是会话隔离、任务调度、进度汇报这些系统工程...
- [Codex CLI 使用技巧 2026：20+ 实战招式与沙箱详解](https://www.heyuan110.com/zh/posts/ai/2026-02-12-codex-cli-mastery-guide/): npm install -g @openai/codex 装好即用。20+ 技巧覆盖模型切换、会话恢复、Option+Enter 换行、关闭沙箱与 MCP 集成，附 Claude Code 对比。
- [OpenClaw 超详细上手教程：小白友好 + 老鸟技巧](https://www.heyuan110.com/zh/posts/ai/2026-02-12-openclaw-usage-tutorial/): OpenClaw 从零到精通的完整教程，涵盖安装、Gateway 启动、TUI 终端、多智能体配置、安全隔离、排障清单等实战内容，适合新手入门和老鸟进阶。
- [Claude-Mem 深度解析：给 Claude Code 装上永久记忆](https://www.heyuan110.com/zh/posts/ai/2026-02-03-claude-mem-deep-dive/): Claude-Mem 是一款让 Claude Code 拥有跨会话持久记忆的插件。本文深入剖析其 Hook 架构、三层搜索、AI 压缩、Endless Mode 等核心技术，并与原生 CLAUDE.md 对比分析。
- [Moltbook 深度解析：AI Agent 专属社交网络的疯狂实验](https://www.heyuan110.com/zh/posts/ai/2026-02-01-moltbook-ai-agent-social-network/): Moltbook 是 2026 年最火的 AI 社交网络实验，150,000 个 AI Agent 在上面自建宗教、辩论意识、组建政府。本文深度解析运作机制、手把手教你加入、剖析 API 架构与安全争议。
- [Superpowers 深度解析：让 Claude Code 变身高级工程师的技](https://www.heyuan110.com/zh/posts/ai/2026-02-01-superpowers-deep-dive/): Superpowers 是 GitHub 上 4 万星的 Claude Code 技能框架，通过 TDD、子代理驱动开发、系统化调试等技能，将 AI 编码助手变成遵循工程纪律的高级开发者。本文深度解析...
- [OpenClaw 作者 Claude Code 工作流：一人 AI 写 10 万](https://www.heyuan110.com/zh/posts/ai/2026-01-31-openclaw-claude-code-workflow/): OpenClaw 作者 Peter Steinberger 的 Claude Code 开发方法论拆解：AGENTS.md 文档驱动、5-10 个 Agent 并行、Spec 驱动构建、日均 600...
- [别再跟 AI 说「帮我做个 XX」：22 个方法论让需求从模糊变清晰](https://www.heyuan110.com/zh/posts/ai/2026-01-31-thinking-methodologies-guide/): 系统梳理 SMART、OKR、MECE、JTBD、RICE、PDCA 等 22 个经典产品方法论与思维模型，覆盖目标定义、问题拆解、需求描述、优先级排序、持续改进五大阶段，每个方法论配真实案例和 AI 协作场景。
- [单元测试报告生成工具全解析：框架对比、覆盖率策略与 AI 实战](https://www.heyuan110.com/zh/posts/ai/2026-01-31-unit-test-report-tools/): 深度解析 2026 年主流单元测试报告生成工具，涵盖 JUnit、PyTest、Allure 等框架对比，以及 AI 辅助测试的实战方法，帮助团队构建高效的测试报告体系。
- [CLAUDE.md vs README.md：一个给 AI 看，一个给人看](https://www.heyuan110.com/zh/posts/ai/2026-01-31-claudemd-vs-readme/): CLAUDE.md 和 README.md 有什么区别？本文结合 OpenClaw（124k Star）的 AGENTS.md 实践和自己项目的真实案例，讲清楚怎么写好这两个文件，让 AI Agent...
- [高频提交不等于失控：日均百次 Commit 的工程方法与落地清单](https://www.heyuan110.com/zh/posts/ai/2026-01-31-high-frequency-commits-strategy/): 高频提交（每天上百次 commit）并不必然导致质量崩盘。本文从原子化提交、Conventional Commits、测试与发布节奏、风险隔离与可回滚等角度，总结一套能兼顾速度与稳定性的工程方法，并给出团队落地清单。
- [OpenClaw 记忆实施策略解析：工具驱动的 RAG 与"按需回忆"](https://www.heyuan110.com/zh/posts/ai/2026-01-31-openclaw-memory-strategy/): 解析 OpenClaw 的记忆实施策略：不把记忆自动注入 prompt，而是通过 memory_search/memory_get 工具按需检索；结合 BM25+向量检索、chunks 表设计与会话增量索引机制。
- [产品思维三板斧：删除优先、质疑一切、亲赴火线](https://www.heyuan110.com/zh/posts/ai/2026-01-30-product-thinking-delete-first/): 从删除优先于优化、质疑惯性思维、亲临用户现场三个维度，拆解一套适合独立开发者和超级个体的产品方法论，附真实踩坑案例。
- [AI 工作流实战手册：从提示词到编程，真正把 AI 用起来](https://www.heyuan110.com/zh/posts/ai/2026-01-30-ai-workflow-real-guide/): 一套经过实战验证的 AI 工作流方法论，涵盖提示词技巧、AI 编程规范、工具选型和企业落地场景，帮你把 AI 从聊天玩具变成生产力工具。
- [Moltbot 是什么？8 万 Star 爆红项目的改名真相与安全避坑指南](https://www.heyuan110.com/zh/posts/ai/2026-01-29-moltbot-deep-dive/): Moltbot（原 Clawdbot/OpenClaw）是 2026 年最火的开源 AI Agent，一周 8 万 Star，能 24 小时自主操控电脑。本文解析 Moltbot 改名原因、核心架构...
- [Claude Code 浏览器自动化怎么选？5 套方案实测对比（2026）](https://www.heyuan110.com/zh/posts/ai/2026-01-28-claude-code-browser-automation/): Browser-use vs Agent Browser vs Playwright CLI vs Playwright MCP vs DevTools MCP，实测 Token 消耗差 10 倍+...
- [Moltbot Wizard 完全指南：打造你的私人 AI 助手](https://www.heyuan110.com/zh/posts/ai/2026-01-28-moltbot-wizard-guide/): Moltbot 是一款开源的个人 AI 助手，通过 Wizard 向导可以快速配置 Gateway、消息渠道和 Skills。本文手把手教你从零开始搭建属于自己的 AI 管家。
- [2026 年 AGI 已经来了：从功能定义到 31 分钟猎头实战](https://www.heyuan110.com/zh/posts/ai/2026-01-26-agi-is-here/): 顶级风投宣称 AGI 已经到来。从功能定义到实战案例，深度解读 AI Agent 如何在 31 分钟内完成猎头工作，以及每 7 个月翻倍的能力增长曲线意味着什么。
- [Claude Code + Remotion：用 AI 对话生成专业视频，程序员](https://www.heyuan110.com/zh/posts/ai/2026-01-26-claude-code-remotion-video/): Remotion 是一个用 React 代码生成视频的开源框架，结合 Claude Code 的 AI 能力，只需用自然语言描述，就能自动生成专业级动画视频。本文手把手教你安装配置，让你 10 分钟上手 AI 视频制作。
- [ClawdBot：海外爆火的个人 AI 管家，30 分钟搭建指南](https://www.heyuan110.com/zh/posts/ai/2026-01-25-clawdbot-personal-ai-assistant/): ClawdBot 是一个开源的 AI Agent 个人助手，可以在你的电脑上 24 小时运行，通过 Telegram、WhatsApp 等聊天工具与你互动，自动处理邮件、日程、提醒等任务。本文手把手教...
- [AI 时代，一个人的 Taste 比以往任何时候都重要](https://www.heyuan110.com/zh/posts/ai/2026-01-23-taste-matters-in-ai-era/): 当AI能够瞬间生成无限内容时，真正的瓶颈不再是创造能力，而是判断什么值得被创造的智慧。本文深度探讨为什么个人品味成为AI时代最稀缺的能力，以及如何培养这种无法被算法复制的核心竞争力。
- [Claude Code 高手都在用的 24 个技巧](https://www.heyuan110.com/zh/posts/ai/2025-01-23-claude-code-commands/): Claude Code 24 个实用技巧合集：从安装启动、快捷键、CLAUDE.md 配置到 MCP 集成、自定义 Slash 命令和 Hooks 自动化，助你把终端 AI 编程助手用到极致
- [Anthropic CEO 彭博社访谈：AI 指数增长、芯片出口与超级智能的未来](https://www.heyuan110.com/zh/posts/ai/2025-01-21-anthropic-ceo-bloomberg-interview/): Anthropic CEO Dario Amodei 在彭博社最新访谈中分享了对 AI 未来的深度见解：从指数级增长的本质，到首席工程师两个月不写代码的现实，再到芯片出口管制和经济影响的预测。本文整理访谈核心观点。
- [Claude Code Skills 排行榜 2026：Top 20 技能包推荐](https://www.heyuan110.com/zh/posts/ai/2026-01-20-claude-code-skills-top20/): 按 GitHub Stars 排的 Claude Code Skills Top 20：官方 anthropics/skills 已 45.1k 星，含 docx/pdf/pptx 等官方技能与社区精选，附安装方法。
- [我的AI开发工作流：从需求到上线](https://www.heyuan110.com/zh/posts/ai/2026-01-19-ai-dev-workflow/): 分享我在软件开发全流程中使用AI工具的实战经验，涵盖需求分析、架构设计、编码实现、测试部署等环节，帮助开发者构建高效的AI辅助开发工作流。
- [Agent Skills：用大白话写程序的时代来了](https://www.heyuan110.com/zh/posts/ai/2026-01-19-agent-skills-new-programming/): 当 Markdown 文件变成可执行的程序，当人人都能用母语编写软件，Agent Skills 正在重新定义什么是编程。这不是简单的提示词，而是新一代的程序。
- [Cursor Agent 编码最佳实践：官方指南完整解读](https://www.heyuan110.com/zh/posts/ai/2026-01-19-cursor-agent-best-practices/): Cursor 官方发布的 Agent 编码最佳实践指南，涵盖计划模式、上下文管理、Rules 与 Skills 配置、测试驱动开发、并行执行等核心技巧，帮助你充分发挥 AI 编程助手的潜力。
- [Claude Code从入门到精通：终端里的全能AI助手完全指南](https://www.heyuan110.com/zh/posts/ai/2026-01-14-claude-code-guide/): 全面介绍 Anthropic 官方 AI 工具 Claude Code，从基础概念到高级用法，涵盖安装配置、日常使用、CLAUDE.md、Skills、Subagent 等核心功能。无论你是开发者、产...
- [Anthropic 发布 Claude Cowork：让 AI 直接操作你的电脑](https://www.heyuan110.com/zh/posts/ai/2026-01-13-claude-cowork/): Anthropic 推出 Claude Cowork 研究预览版，这是一个可以直接访问本地文件、运行代码并与第三方应用集成的桌面 AI 代理，目前仅限 macOS 上的 Max 订阅用户使用。
- [Vercel 发布 Agent Browser：专为 AI 代理打造的浏览器自动](https://www.heyuan110.com/zh/posts/ai/2026-01-13-vercel-agent-browser/): Vercel 开源了 Agent Browser，一个专为 AI 代理设计的浏览器自动化 CLI 工具，采用 Rust + Node.js 架构，支持快照驱动的交互模式
- [小白也能解锁 Claude Code 的秘密武器：Skills](https://www.heyuan110.com/zh/posts/ai/2026-01-12-claudecode-skill-patterns/): 该如何让 AI 学习到我的能力呢？Skills 彻底改变了我对 AI 协作的看法
- [Claude Code 记忆术：一个文件让 AI 永远记住你是谁](https://www.heyuan110.com/zh/posts/ai/2026-01-12-claudemd-memory-guide/): 手把手教你配置 CLAUDE.md，让 Claude Code 记住你的开发环境、代码风格和工作习惯，告别重复自我介绍
- [Claude Code Skill 完全指南：让 AI 学会你的工作流程](https://www.heyuan110.com/zh/posts/ai/2026-01-08-claudecode-skill-guide/): 手把手教你创建 Claude Code Skill，附带财务分析、面试问题生成、亚马逊 Listing 优化三个实战案例
- [Claude Code创始人的几个最佳实践](https://www.heyuan110.com/zh/posts/ai/2026-01-06-claudecode-best-practices/): Claude Code 创始人在 X 上分享的几条使用心得，关于并行工作、模型选择、CLAUDE.md 维护和验证循环
- [Skill与MCP的区别：两种扩展AI能力的方式](https://www.heyuan110.com/zh/posts/ai/2026-01-06-skillmcp/): 深入对比 Claude Code 中 Skill 和 MCP 两种扩展 AI Agent 能力的方式，分析它们的设计理念、工作原理和适用场景
- [Claude Code的Skill与SubAgent区别 ](https://www.heyuan110.com/zh/posts/ai/2025-12-26-claudecode-skillsubagent/): 深入对比Claude Code中Skill和SubAgent两种扩展AI Agent能力的方式，分析上下文管理差异和各自适用场景
- [向量数据库到底解决了什么问题？从原理到工程实践一次讲清](https://www.heyuan110.com/zh/posts/ai/2025-03-11-vectordatabase/): 从原理到工程实践，讲清楚向量数据库为什么火、怎么用、怎么选，以及实际落地中的坑
- [全面解析FaceFusion：从原理到实战](https://www.heyuan110.com/zh/posts/ai/2024-10-16-facefusion/): FaceFusion 换脸工具的完整使用指南，包括硬件选型、环境搭建、参数调优和常见问题解决

## DATAWAREHOUSE（中文） — 1


- [Amazon Redshift 性能优化指南：VACUUM、ANALYZE 与运](https://www.heyuan110.com/zh/posts/datawarehouse/2018-08-09-dw-redshift/): Amazon Redshift 数据仓库性能优化完全指南，详解 VACUUM 六种类型（FULL、DELETE ONLY、SORT ONLY、REINDEX、RECLUSTER、BOOST）、ANAL...

## DOCKER（中文） — 6


- [Docker Compose 教程：docker-compose.yml 配置详](https://www.heyuan110.com/zh/posts/docker/2026-01-24-docker-compose-yml-explained/): 最全 Docker Compose 教程：逐字段讲透 docker-compose.yml 的 services、volumes、networks、healthcheck 配置，附 WordPress...
- [Cloudflare Workers 完全指南：从入门到实战的边缘计算部署手册](https://www.heyuan110.com/zh/posts/docker/2026-01-23-cloudflare-workers-guide/): Cloudflare Workers 是部署无服务器应用的最佳选择之一。本文详解 Workers 的核心概念、免费额度、适用场景、开发部署流程，并提供 API 代理、短链接服务等实战案例，助你快速掌握边缘计算开发。
- [Docker Compose 完全指南（2026）：安装、docker-comp](https://www.heyuan110.com/zh/posts/docker/2026-01-19-docker-compose-complete-guide/): 从 Docker 基础到 Docker Compose 实战：安装配置、compose.yml 核心字段、WordPress/Node+MySQL 案例与生产环境最佳实践。
- [Docker 常用命令速查（2026更新）：镜像、容器、网络、数据卷一页掌握](https://www.heyuan110.com/zh/posts/docker/2019-11-14-docker-commands/): 高频 Docker 命令速查：pull/run/exec/logs、镜像清理、网络管理、数据卷挂载，附 MySQL/Nginx/Redis 常用启动示例。
- [Nexus3 搭建 Docker 私有镜像仓库：企业级 Registry 完整教](https://www.heyuan110.com/zh/posts/docker/2019-06-12-next3-dockerhub/): 使用 Nexus3 搭建企业级 Docker 私有镜像仓库完整教程，详解 hosted、proxy、group 三种仓库类型配置，实现镜像的推送和拉取，附 Ubuntu 客户端 insecure-re...
- [Docker 入门教程：镜像、容器、仓库核心概念与 Dockerfile 详解](https://www.heyuan110.com/zh/posts/docker/2019-05-13-learn-docker/): Docker 零基础入门教程，详解镜像、容器、仓库三大核心概念，深入理解分层存储原理，Dockerfile 指令全解析（FROM、RUN、COPY、CMD、ENTRYPOINT 等），附私有仓库搭建和数据卷实战

## ELASTICSEARCH（中文） — 3


- [AWS EKK 日志系统搭建：Elasticsearch + Kinesis +](https://www.heyuan110.com/zh/posts/elasticsearch/2018-09-12-log-ekk/): 基于 AWS 托管服务搭建 EKK 日志系统完整教程，使用 Amazon Elasticsearch Service、Kinesis 和 Kibana 实现 Nginx 日志收集分析，比 ELK 更易运维
- [Elasticsearch 入门教程：索引、文档、查询 API 核心概念详解](https://www.heyuan110.com/zh/posts/elasticsearch/2018-09-12-elasticsearch/): Elasticsearch 入门教程，理解索引、类型、文档等核心概念（类比关系型数据库），掌握 _cat、_search 等常用查询命令和 REST API 操作
- [ELK 日志系统搭建指南：Elasticsearch + Logstash +](https://www.heyuan110.com/zh/posts/elasticsearch/2018-09-11-log-elk/): 基于 Elasticsearch、Logstash、Kibana 和 Kafka 搭建企业级 ELK 日志收集分析系统完整教程，对比多种架构模式，附 Filebeat 配置

## GO（中文） — 1


- [Go 语言入门教程：Golang 基础语法与数据类型详解](https://www.heyuan110.com/zh/posts/go/talk/2019-01-25-learn-go/): Go 语言零基础入门教程，介绍 Golang 特点（部署简单、并发性好、执行性能好），详解基础语法结构、包声明、函数定义、变量使用、数据类型（布尔、整型、浮点、字符串）等核心概念

## JAVA（中文） — 2


- [Java 入门教程（二）：Arrays 工具类、集合框架与 I/O 操作详解](https://www.heyuan110.com/zh/posts/java/2024-05-13-learn-java2/): Java 入门教程第二篇，详解 Arrays 工具类常用方法（toString、copyOf、sort、setAll），数组遍历、拷贝、排序操作，以及集合框架和 I/O 流技术基础
- [Java 入门教程（一）：JDK/JRE/JVM 概念解析与 IDEA 开发环境](https://www.heyuan110.com/zh/posts/java/2024-05-05-learn-java/): Java 零基础入门教程第一篇，涵盖 Java 发展历史、SE/EE/ME 技术体系，详解 JDK、JRE、JVM 概念区别，IDEA 开发环境配置和常用快捷键

## LINUX（中文） — 14


- [fish shell 4.6 实测：最舒服的交互 shell，但别当默认](https://www.heyuan110.com/zh/posts/linux/2026-04-18-fish-shell-rust-2026/): fish shell 4.6（2026 年 3 月发布）实测九个月结论：Rust 重写 2600+ commits 却没让 shell 变快，POSIX 不兼容还会坑 Claude Code 等 AI...
- [把公司 Mac Mini 变成你的私人 VPN：5 种方案实测对比](https://www.heyuan110.com/zh/posts/linux/2026-04-11-remote-access-mac-mini-vpn/): 亲测 5 种远程访问公司 Mac Mini 的方案 —— Tailscale、WireGuard、ZeroTier、Cloudflare Tunnel、frp，含完整搭建教程、国内可用性实测和踩坑记录。
- [Shell 特殊变量完全指南：$$、$?、$@、$# 等用法详解](https://www.heyuan110.com/zh/posts/linux/2019-05-13-linux-shell-vars/): Bash Shell 特殊变量完全指南，详解 $$、$!、$?、$-、$*、$@、$#、$0 等变量的含义和实际用法。包含 $* 与 $@ 的区别对比、退出状态码说明、完整示例代码和最佳实践。
- [AWS CLI 完全指南：安装配置、S3/EC2 常用命令与权限排错](https://www.heyuan110.com/zh/posts/linux/2020-07-04-aws-cli/): AWS CLI 从零上手：安装、aws configure、S3/EC2/IAM 高频命令、常见报错与权限问题排查，适合运维与开发日常使用。
- [Nginx HTTPS 配置完全指南：自签名证书、混合站点与反向代理踩坑](https://www.heyuan110.com/zh/posts/linux/nginx/nginx-https-complete-guide/): Nginx HTTPS 配置实战合集：OpenSSL 自签名 SSL 证书生成、部分页面 HTTPS 混合站点配置、反向代理 AWS ELB 时 DNS 动态解析踩坑与解决方案
- [curl 命令详解（2026）：GET/POST 请求、文件传输与 API 调试](https://www.heyuan110.com/zh/posts/linux/2020-06-29-curl/): curl 命令完整教程：GET/POST/PUT/DELETE 请求、JSON 数据发送、文件上传下载、鉴权认证、代理配置与耗时分析，一文速查全部用法。
- [traceroute 命令详解（2026）：路由追踪与网络故障排查实战指南](https://www.heyuan110.com/zh/posts/linux/2020-06-28-traceroute/): traceroute/tracert 命令完整教程：TTL 原理、参数用法、输出解读、故障排查实战，覆盖 Linux/macOS/Windows 三平台。
- [Linux/macOS 常用命令速查手册（持续更新）](https://www.heyuan110.com/zh/posts/linux/2020-03-19-linux-mac-commands/): Linux 和 macOS 常用命令速查手册，涵盖文件操作、网络调试、进程管理、文本处理、磁盘管理等日常运维开发必备命令，附带实用示例
- [Ubuntu 搭建 Jira + Confluence 完整教程](https://www.heyuan110.com/zh/posts/linux/2019-04-15-jira-confluence-install/): Ubuntu 16.04 搭建 Jira 项目管理和 Confluence 知识库完整教程，包含 MySQL 配置、破解激活、账号整合、数据备份恢复和安全漏洞修复
- [Linux 挖矿病毒 kerberods 入侵排查与清理实战](https://www.heyuan110.com/zh/posts/linux/2019-04-12-linux-kerberods/): 记录 Linux 服务器遭受 kerberods 挖矿病毒入侵的完整排查和处理过程，CPU 持续 100%、top 只显示 1 个 CPU 的诡异现象分析，附清理脚本和 Confluence 漏洞修复
- [Jenkins + AWS CodeDeploy + AutoScaling 持](https://www.heyuan110.com/zh/posts/linux/2018-11-20-jenkins-codedeploy-autoscaling/): 基于 Jenkins、AWS CodeDeploy 和 AutoScaling 搭建 CI/CD 持续集成流水线完整教程，详解蓝绿部署、就地部署原理，AppSpec 配置和生命周期钩子
- [Supervisor 教程（2026）：安装配置、常见问题与 Superviso](https://www.heyuan110.com/zh/posts/linux/2018-10-07-supervisor/): 面向 Linux 运维与开发者的 Supervisor 实战指南：安装配置、进程托管、常见故障排查，以及与 systemd/PM2 的选型对比。
- [IP 地址与 CIDR 详解：网络基础知识必备指南](https://www.heyuan110.com/zh/posts/linux/2018-10-06-ip-cidr/): IP 地址与 CIDR 完整教程，详解 IPv4 地址结构、A/B/C/D/E 类地址划分、子网掩码计算、CIDR 无类别域间路由。附子网速查表，网络工程师和运维必备的网络基础知识。
- [Oh My Zsh 安装配置指南：打造高效终端环境](https://www.heyuan110.com/zh/posts/linux/2015-06-17-shell-zsh/): Oh My Zsh 完整安装配置教程，涵盖 Zsh 安装、Oh My Zsh 配置、Powerlevel10k 主题美化、zsh-autosuggestions 和 zsh-syntax-highli...

## MACOS（中文） — 2


- [终端工具推荐：23 款终端模拟器跨平台实测对比](https://www.heyuan110.com/zh/posts/macos/2025-01-22-terminal-tools-guide/): iTerm2、Warp、Ghostty、Windows Terminal、Tabby 等 23 款终端对比，覆盖 macOS/Windows/Linux 三平台，附 Windows shell 与 SSH 客户端选型。
- [Charles 抓包教程（2026）：HTTP/HTTPS 拦截、Mock 调试](https://www.heyuan110.com/zh/posts/macos/charles/): Charles 抓包实战指南：HTTP/HTTPS 请求拦截、Mock 数据、网络限速与断点调试，补充 2026 证书安装与常见问题排查。

## MANAGEMENT（中文） — 1


- [刘澜领导力十项修炼学习总结：领导力本质与管理者成长](https://www.heyuan110.com/zh/posts/management/2019-08-19-liulan-management/): 刘澜老师领导力课程学习笔记，解析领导力的本质（责任而非职位）、魅力的副作用，以及领导力十项修炼口诀，适合技术管理者和团队 Leader 阅读

## MIDDLEWARE（中文） — 3


- [Redis 从入门到精通：安装配置与实战指南](https://www.heyuan110.com/zh/posts/middleware/2026-01-22-redis-complete-guide/): Redis 完全指南，详解 macOS 和 Windows 安装方法、五大数据类型、常用命令、持久化机制、主从复制与集群配置。从零基础到精通 Redis 的最佳学习路径。
- [RabbitMQ 入门教程：AMQP 协议、Exchange 交换机与消息投递详](https://www.heyuan110.com/zh/posts/middleware/2019-08-02-mq-rabbitmq/): RabbitMQ 消息队列入门教程，详解 AMQP 协议原理、四种 Exchange 交换机类型（Direct、Fanout、Topic、Headers）、队列绑定和消息投递工作流程
- [消息队列 MQ 技术选型指南：核心概念与分布式系统应用场景](https://www.heyuan110.com/zh/posts/middleware/2019-07-31-mq/): 消息队列技术选型完整指南，解析 MQ 核心概念（Producer、Broker、Consumer）、架构模型，以及在分布式系统中的异步处理、削峰填谷、系统解耦等应用场景

## MYSQL（中文） — 2


- [MySQL SQL 完全指南：从小白到高手的进阶之路（含面试宝典）](https://www.heyuan110.com/zh/posts/mysql/2026-01-23-mysql-sql-complete-guide/): 全面掌握 MySQL SQL 核心知识：从基础语法到高级特性，涵盖索引原理、事务隔离、锁机制、查询优化。包含大量实战案例和高频面试题，助你快速从 SQL 小白进阶为数据库高手。
- [MySQL EXPLAIN 执行计划详解：从入门到精通的 SQL 性能分析指南](https://www.heyuan110.com/zh/posts/mysql/2019-09-06-mysql-explain/): 深入解析 MySQL EXPLAIN 执行计划的 12 个关键字段，掌握 type 访问类型、key 索引选择、Extra 额外信息的含义，学会通过执行计划分析和优化 SQL 查询性能。

## PYTHON（中文） — 3


- [uv、conda、pyenv 到底怎么选？2026 实测对比与迁移指南](https://www.heyuan110.com/zh/posts/python/2026-07-27-uv-vs-conda-vs-pyenv/): 同一台 Mac 实测三者：uv 热缓存 0.10 秒装完三件套，比 pip 快 52 倍；conda 环境占 275MB 还弹出付费条款。附清华镜像配置、conda/pyenv 迁移路径与命令对照速查表。
- [uv 完全指南：一个工具干掉 pip + conda + pyenv](https://www.heyuan110.com/zh/posts/python/2026-04-10-uv-python-package-manager/): uv 是 Rust 写的 Python 包管理器，比 pip 快 10-100 倍：实测装 JupyterLab pip 要 21.4 秒、uv 只需 2.6 秒，2026 年月下载量已达 7500...
- [Conda 完全指南：Python 多版本环境管理从入门到精通](https://www.heyuan110.com/zh/posts/python/2020-01-11-python-conda/): Conda 是 Python 开发者必备的环境管理工具，本文详解 Conda 与 pip 的区别、Anaconda 与 Miniconda 的选择、环境创建与管理、常用命令及最佳实践，帮助你轻松管理...

## TYPESCRIPT（中文） — 1


- [TypeScript 完全指南：从入门到精通的进阶手册](https://www.heyuan110.com/zh/posts/typescript/2026-01-23-typescript-complete-guide/): TypeScript 全面教程，涵盖基础语法、高级类型、泛型、类型体操、工具类型、最佳实践。适合新手入门和面试复习，附带大量实战代码示例。2025年最新版本。