AI Developer Tools归档

Cloudflare’s Open-Source security-audit Skill: Have Claude Code or Codex Audit Your Codebase, How the Six Phases Work and What It Costs

AI Developer Tools Cloudflare’s Open-Source security-audit Skill: Have Claude Code or Codex Audit Your Codebase, How the Six Phases Work and What It Costs

security-audit is an open-source coding agent skill from Cloudflare’s security team (MIT, about 34K stars) and the starting point of its internal vulnerability-hunting system. Install it in Claude Code or Codex, say “security audit this codebase”, and it runs reconnaissance, hunting, adversarial validation, structured output, independent review and reporting. We tested the install and bundled tests and explain the two modes, three tiers, sandboxing and cost.

ai-memory: Switch from Claude Code to Codex Mid-Task Without Re-Explaining Your Project, with Open-Source Long-Term Memory for AI Coding Agents

AI Agents ai-memory: Switch from Claude Code to Codex Mid-Task Without Re-Explaining Your Project, with Open-Source Long-Term Memory for AI Coding Agents

Ran out of Claude Code quota and switched to Codex, or moved to another machine, and the AI forgot everything? ai-memory uses hooks to record sessions automatically and curates them into a git-managed Markdown wiki that 20+ coding agents share, with automatic handoff when you switch. MIT-licensed, no API key needed by default. This post covers installation, daily use, privacy limits and tips for non-English projects.

AI Developer Tools DeepSeek V4.1 Flash 1017 Update FAQ: Benchmark Gains, Pricing, the Vision API and Claude Code Setup

On October 17, DeepSeek updated V4.1 Flash to the 1017 release across the web app, mobile app and API. The model name is still deepseek-flash and the price is unchanged. This FAQ covers the benchmark changes, peak and off-peak pricing, how to send images to the API and its limits, the thinking-mode switch, Claude Code configuration, and whether you can run the open weights at home.

AI Agents PilotDeck 2.0: An Open-Source AI Agent Workspace from Tsinghua and ModelBest, with One-Command Install and Word/Excel/PPT Skills

PilotDeck 2.0 (OpenBMB, ModelBest and THUNLP) brings one-command install, white-box memory, smart routing and Word/Excel/PPT Skills. About 4,495 stars, AGPL-3.0.

AI Agents Magpie: Switch Models for Claude Code and Codex from Your Menu Bar, with a Local Gateway for DeepSeek and Kimi

Magpie is an MIT-licensed local model gateway for coding agents: switch models from the menu bar while 127.0.0.1:3425 translates between OpenAI, Anthropic and Gemini APIs, with multi-account routing. About 13k stars.

AI Developer Tools Strata: Run Qwen3.8-Flash-Next on a Gaming PC and Serve an OpenAI/Anthropic API to Your Agents (Check the Hardware Bar First)

Strata is an MIT-licensed local inference engine that runs Qwen3.8-Flash-Next on consumer GPUs and serves a localhost API for coding agents. About 25k stars, but it needs 12 GB+ of VRAM and lots of RAM. Here’s the real cost.

AI Agents Mobile MCP: Let Claude Code and Cursor Control iOS and Android Simulators and Real Devices

Mobile MCP is an Apache-2.0 open-source MCP server that lets Claude Code, Cursor and other agents drive iOS and Android simulators and real devices with one toolset. Accessibility-first, with optional cloud devices.

AI Agents OpenCode 2.0: Local Service, Code Mode and What Breaks When You Migrate

OpenCode 2.0 brings a background local service, Code Mode, and rebuilt plugins and SDK. Here are the new features, the breaking changes, and the config-precedence security issue discussed on HN and reportedly reverted.