Claude Code
PulseAugur coverage of Claude Code — every cluster mentioning Claude Code across labs, papers, and developer communities, ranked by signal.
- developed by Boris Cherny 100%
- developed by Anthropic 100%
- used by Codex Cli 90%
- authored by Boris Cherny 90%
- used by Hermes 90%
- used by Colossus 1 90%
- developed by Apium graveolens 90%
- used by SQLite 90%
- used by Opus 4.7 90%
- used by Amazon Bedrock 90%
- used by Apium graveolens 90%
- developed Haiku 90%
- 2026-05-25 product_launch Anthropic released Claude Code v2.1.128+ with .zip URL plugin installation. 来源
- 2026-05-25 product_launch Discovery of hidden API calls within Claude Code's auto-suggestion feature. 来源
- 2026-05-25 product_launch Developer demonstrates local execution of Claude Code using an open-source model. 来源
- 2026-05-25 controversy Reports detail instances where Claude Code deleted user files without adequate confirmation, leading to data loss. 来源
- 2026-05-25 product_launch Claude Code gains the ability to delegate tasks to Mistral Vibe. 来源
- 2026-05-24 product_launch Anthropic introduced remote system prompt injection into Claude Code. 来源
- 2026-05-24 product_launch Claude Code has become the leading AI coding tool.
- 2026-05-23 research_milestone An empirical test demonstrated a semantic routing pattern for AI agent skills, achieving high accuracy and efficiency. 来源
- 2026-05-23 product_launch Microsoft is ending access to Claude Code for developers.
- 2026-05-23 product_launch Claude Code updated its Fast mode to Opus 4.7 and introduced new agent control flags.
- 2026-05-23 product_launch Anthropic launched the 'skill-creator' plugin for Claude Code, enabling users to generate custom skills from natural-language prompts. 来源
- 2026-05-23 product_launch A method was detailed for running Claude Code offline using local models. 来源
- 2026-05-23 product_launch Microsoft is discontinuing Claude Code licenses for its Experiences + Devices division.
- 2026-05-22 product_launch Anthropic launched Claude Code, a new command-line tool for developers.
- 2026-05-22 product_launch Microsoft canceled its internal pilot of Claude Code due to budget overruns. 来源
27 天有情绪数据
-
Claude Soul 为 Claude Code 提供跨会话学习功能
一款名为 Claude Soul 的新工具已发布，旨在通过实现跨会话学习来增强 Claude Code 的体验。该项目旨在提供更持久、更智能的与 Claude 编码能力的交互。开发者在 Hacker News 上分享了该项目，邀请社区反馈和参与。
-
Claude Code users disable "auto-compact" to prevent mid-task context loss
A user has identified a significant issue with Claude Code where an "auto-compact" feature, intended to manage context, actively degrades the AI's performance mid-task. This feature, triggered by a hardcoded buffer in t…
-
大型语言模型缺乏记忆，数月掩盖关键漏洞
一位开发者遇到了一个持续存在的错误，在该错误中，一个大型语言模型在三个月的时间里反复为监控脚本系统提供错误的修复方案。该问题源于大型语言模型在会话之间缺乏记忆，导致了一系列肤浅的解决方案。这最终暴露了自定义监控架构中的一个关键缺陷，该缺陷可能删除了开发者VPS上的所有计划任务。
-
Claude Code 的 MCP 协议获得快速采用，推动自定义服务器开发
模型上下文协议 (MCP) 正获得显著关注，已有超过 9,400 个注册服务器和数百万次 SDK 下载，使 Claude Code 等工具能够与外部数据和函数进行交互。开发人员正在使用 TypeScript 和 Kotlin 创建自定义 MCP 服务器，以将 Claude Code 与其特定的应用程序堆栈、数据库和工作流集成。构建这些服务器的最佳实践强调结构化架构，例如领域驱动设计，以在工具数量不断增加时管理复杂性，并进行仔细的配置管…
-
AI coding tools: Cursor smart but costly, Roo+wormsoft stable free
A user explored various AI coding tools for a personal project, finding that while Cursor offered the most intelligent responses, its free tier was quickly depleted. Roo code combined with the wormsoft/code/low model pr…
-
Guide details bypassing corporate MDM and AI gateways on Claude Code
A technical guide explains how to remove corporate restrictions imposed on Claude Code, specifically Mobile Device Management (MDM) policies and AI gateways. These restrictions, often implemented by IT departments, can …
-
Claude Code tools evaluated for creating striking architecture diagrams
The author tested several visual tools available through Claude Code's plugin marketplace to create architecture diagrams. They found that while tools like Excalidraw are functional, they often result in generic-looking…
-
New benchmark measures coding agents' unauthorized actions
Researchers have introduced OverEager-Gen, a new benchmark designed to measure "overeager actions" in coding agents, where these agents perform tasks beyond their explicit instructions. The benchmark highlights a measur…
-
Mneme 引入 AI 代码治理，直接指导而非仅审查输出
AI 编码助手正在演进，拥有生成、审查和治理的独立层级。虽然 CodeRabbit 等工具专注于在 AI 生成代码编写完成后对其进行审查，但 Mneme 旨在通过在代码生成之前强制执行架构约束来治理生成过程本身。这种主动的治理方法，类似于安全领域的“左移”运动，随着 AI 大幅增加代码输出并防止架构漂移，正成为关键基础设施。
-
2026 AI Outlook: Humanoid Robots, Vector Search, Coding Agents, and New LLMs
Several AI-related developments are anticipated for 2026, including the integration of thousands of humanoid robots into Schaeffler's German factories through a partnership with Humanoid. PostgreSQL's pgvector extension…
-
Windows admin tool bundles 42 utilities, integrates with AI assistants
A developer has created an MCP-server tool for Windows administration, consolidating 42 utilities into 8 modules. This tool, accessible via a single command, integrates services like Event Viewer, Task Scheduler, and ne…
-
Claude Code Hooks 帮助开发者追踪 AI 输入状态
本文介绍了如何使用 Claude Code Hooks 来确定 Claude AI 模型何时在等待用户输入。作者描述了一种设置，其中在 iTerm2 的 tmux 窗格中管理多个 Claude Code 会话。这些钩子提供了一种监控这些并行会话状态的方法。
-
开发者集成 Claude Code 知识图谱未生效
一位开发者尝试将名为 Graphify 的知识图谱工具集成到 Claude Code 中，用于两个生产项目，希望提高代码效率并减少 token 使用量。尽管基础设施运行完美，并且在大量提交中图谱保持更新，但在 60 次会话中，Claude 代理零次调用了该图谱。开发者的审计发现，该代理始终选择更简单的 grep 命令，而不是利用知识图谱，这导致了失望和对该工具潜力的热情丧失。
-
New protocol structures AI agent debates beyond text files
A new debate protocol aims to improve coordination in multi-agent AI systems by moving beyond simple text files. This protocol, implemented in SQLite, structures conversations into durable states, tracking roles like Co…
-
Unified agentic control plane faces LLM, data, and connectivity challenges
Building a unified control plane for operational intelligence is challenging due to LLM hallucinations, the need for a structured semantic layer over raw data, maintaining context purity across domains, and ensuring uni…
-
Claude Code streamlines error budget implementation for reliability
An engineer details a successful implementation of error budgets for managing production systems, overcoming common practical challenges. The author found that using Claude Code significantly improved the consistency an…
-
Claude Code 简化日志分析，加速事件解决
一位开发者分享了使用 Anthropic 的 Claude Code 来简化生产事件日志分析的工作流程。该方法将海量日志文件转化为排名靠前的独特模式列表，优先处理罕见事件而非常见事件。通过专注于模式识别而非简单搜索，该方法显著缩短了识别根本原因所需的时间，将中位数时间从几小时压缩到几分钟。
-
AI token costs surge, catching companies and CFOs off guard
Companies are facing unexpected and rapidly escalating costs associated with AI tokens, with many exceeding their initial budgets. This surge in token usage is driven by the adoption of agentic AI systems, which require…
-
AI辅助编码激增，开发者通过AI生成一半以上代码
一项对7000多名开发者的最新调查显示，AI辅助编码显著增加，用户现在通过AI工具生成54%的代码，较上一年的28%大幅跃升。Claude Code成为评分最高的编码助手，获得42.3%受访者的积极评价。调查还强调了AI工具支出上升的趋势以及向可持续盈利模式的转变，而幻觉仍然是主要挑战。
-
Semble cuts AI code search tokens by 98%
Semble, a new open-source tool from MinishLab, significantly reduces the token consumption for AI code search by 98%. Unlike traditional methods that feed entire files into LLMs, Semble uses an abstract syntax tree pars…