#AI编程
9352 related articles
Claude Code Advanced Guide: MCP Integration, Writing Skills, and Subagent Code Review
A Bilibili tech sharing session systematically explains advanced Claude Code usage: installation and integration, built-in commands, MCP tool integration, writing Skills, subagent code review, and CLAUDE.md configuration, while addressing real-world account ban concerns.
Claude Writes 90% of the Code—So Who Reviews It? Inside the Seven-Layer Review System
About 90% of Claude Code is AI-generated. Anthropic rebuilt code review with seven layers—AI self-checks, risk-tiered human approval, Merge Queue, and fast rollback—doubling PRs while cutting rollbacks over 40%.
Where Did the Once-Viral 'Crayfish' Go? Code Teardown Reveals It's Just a Subset of Claude Code
Where did the viral multi-agent framework 'Crayfish' go? A Bilibili creator's code teardown reveals its core is just a subset of Claude Code—task decomposition and interface control now absorbed into mainstream dev tools. Plus: the real magnitude of AI coding efficiency gains, the moving boundary of AI's definition, and social solutions to multi-agent permission security.
Freshman Year: Install Codex and Claude Code First? An AI Skills Starter Guide
Bilibili creator advises freshmen to install Codex and Claude Code early, sharing a three-dimensional guide on education, tech stack, and projects for non-elite undergrads entering Agent development, covering Python, RAG, and prompt engineering.
Claude Ports Rust Code: 9,628 of 10,000 Results Don't Match — Where's the Problem?
Bilibili creator tests Claude Code porting a Python function to Rust: zero compile warnings, but 9,628 of 10,000 test cases mismatch. Three traps the compiler can't see, and why you shouldn't add features during a port.
Replit in Action: Cutting Customer Issue Resolution Time by 60% in 6 Months
Travis's team used Replit to build a triage app, cutting customer issue resolution time by 60% in 6 months. This article analyzes the development model, how the triage app works, and implications for customer support teams.
Anthropic Restricts Claude Code to Prevent Game Cloning, Users Seek Workarounds
Anthropic confirms new restrictions on Claude Code to prevent game cloning. Community discussion suggests they may target only the Opus 5.5 model with Claude Code, and developers are exploring workarounds such as switching frameworks.
@ai-sdk/svelte 4.0.321 Patch Released: The AI SDK Ecosystem Keeps Iterating
@ai-sdk/svelte releases patch version 4.0.321, synced with core package ai@6.0.321. This article covers the technical background, upgrade advice, and what the AI SDK ecosystem's continued Svelte support means.
Qodo Injects Context into Codex for More Precise AI Code Review
Qodo injects company knowledge and coding standards into Codex, giving AI code review organizational context to catch P0 issues across repos before a PR. This article analyzes the approach and its practical impact.
AI SDK Svelte 5.0.160 Patch Release: Behind the Dependency Chain Update
AI SDK's Svelte adapter package releases patch 5.0.160, syncing the ai core package to 7.0.160. This article explains the dependency update, its impact on Svelte developers, and upgrade advice.
@ai-sdk/tui 1.0.161 Released: Dependency Updates and the Evolution of the AI SDK Ecosystem
@ai-sdk/tui releases version 1.0.161, updating dependencies and syncing the ai package to 7.0.160, reflecting the AI SDK ecosystem's continuous iteration and automated release practices.
AI SDK Tool bash 1.0.14 Release Update Analysis
AI SDK tool bash 1.0.14 released, updating dependencies including ai@7.0.160. This patch update improves compatibility.
AI SDK Toolkit Fixes Network Request Error Handling
AI SDK releases @ai-sdk/tool-web-fetch 1.0.10, fixing the issue where an invalid Location header in redirects threw an exception; it now returns a network_error result, improving tool robustness.
AI SDK Vue 3.0.321 Patch Released: Synced with ai@6.0.321
AI SDK Vue adapter package releases patch 3.0.321, synced with core package ai@6.0.321. A routine version alignment update with no new APIs. This article covers the patch contents, upgrade notes, and AI SDK multi-package version management best practices.
@ai-sdk/workflow 2.0.92 Released: Dependency Upgrades Strengthen Workflow Stability
@ai-sdk/workflow releases patch version 2.0.92, updating multiple dependencies and upgrading the ai package to 7.0.160, with 13 commits. This article explains the changes, their impact on developers, and upgrade considerations.
AI SDK Vue 4.0.160 Released: A Patch Update and Dependency Upgrade
@ai-sdk/vue 4.0.160 released: a patch update syncing dependency ai@7.0.160. This article explains what the version means, its impact on Vue developers, and AI SDK's release cadence.
AI SDK workflow-harness 1.0.171 Patch Released
AI SDK's @ai-sdk/workflow-harness releases patch 1.0.171, synchronously updating @ai-sdk/harness with 13 commits. This article explains the patch changes, version sync strategy, and upgrade advice for developers.
AI SDK xAI 5.0.36 Fix: No More False 'other', Correctly Returns 'length'
AI SDK's xAI provider releases the 5.0.36 patch, fixing the finish reason mapping in Responses mode: truncated responses due to max_output_tokens now correctly return length instead of other, making truncation handling and observability logic actually work.
Anthropic Tightens Claude Code: No Replicating Classic Games
Anthropic updated Claude Code's system prompt to ban replicating existing video games, highlighting copyright concerns, effects on retro gaming, and AI coding tool regulation.
Vercel's New Defense Line: Blocking Malicious npm Packages Before and During Installation
Vercel introduces an npm package security interception mechanism that checks a blocklist before and during dependency installation to stop malicious packages from running install scripts, automatically applied to new builds on all plans with zero configuration.