OAuth
PulseAugur coverage of OAuth — every cluster mentioning OAuth across labs, papers, and developer communities, ranked by signal.
3 天有情绪数据
-
WorkOS 推出 auth.md 以标准化 AI 代理 Web 应用注册
WorkOS 推出了新的开放协议 auth.md,旨在简化 AI 代理向 Web 应用注册的方式。该协议利用现有的 OAuth 标准实现自动化身份验证,无需人工干预。Auth.md 定义了两种不同的注册流程:“代理已验证”(Agent Verified),其中代理的身份提供商证明用户身份;以及“用户声明”(User Claimed),使用通过电子邮件发送的一次性代码。目标是通过提供结构化且可审核的凭证颁发和撤销方法,来标准化代理对企业系统的访问。
-
AI代理支出需要通话前预算执行
需要一种新的方法来管理AI代理的支出,因为当前的令牌计数器和可观测性工具不足。提出的解决方案涉及实施一个通话前预算执行系统,类似于Stripe等服务使用的支付授权和捕获机制。该系统将在代理调用前预留资金,之后提交实际成本,并为每次交易提供可审计的签名收据,以防止成本失控。
-
FBI警告Kali365网络钓鱼工具窃取Microsoft OAuth令牌
FBI已发出警告,指出Kali365网络钓鱼工具正在大规模有效地窃取Microsoft OAuth令牌。该复杂工具通过诱骗用户提供凭据来绕过多因素身份验证。该网络钓鱼工具以Microsoft 365账户为目标,授予攻击者广泛的访问权限。
-
MCP 通过 RFC 9207 签发者验证增强 OAuth 安全性
模型上下文协议 (MCP) 已更新其授权流程以符合 RFC 9207,增强了针对 OAuth 混淆攻击的安全性。此更改要求授权服务器在其响应中包含 `iss` 参数,客户端随后必须根据最初记录的签发者验证此参数。这种结构性防御可防止攻击者欺骗客户端使用与错误身份提供商相关的授权码,这是先前基于会话的方法无法完全解决的漏洞。
-
AI 代理需要面向用户的 OAuth 以实现安全访问
AI 代理需要比简单的 API 密钥更强大的身份验证方法,才能安全地访问用户特定数据并执行操作。面向用户的 OAuth 通过允许单个用户授予代理有范围的、可撤销的权限来解决此问题,从而确保明确的同意并实现精细控制。这种方法对于建立信任和扩展 AI 代理应用程序至关重要,将它们从基本原型提升到企业级解决方案。
-
Codens simplifies setup with new PyPI package, reducing onboarding friction
The author describes the development of a new PyPI package, "codens-mcp", designed to streamline the onboarding process for their suite of AI tools. Previously, users had to configure five separate server entries and ru…
-
Stripe Link wallet now lets AI agents shop and pay with user approval
Stripe has launched an updated version of its digital wallet, Link, which now allows autonomous AI agents to make purchases and manage payments on behalf of users. This new functionality enables agents to handle tasks l…
-
Cloudflare enables AI agents to create accounts, buy domains, and deploy code
Cloudflare has introduced a new protocol, co-developed with Stripe, that allows AI agents to autonomously create accounts, register domains, and deploy applications. This system enables agents to perform all necessary s…
-
AI coding tools require clear intent to bridge gap between running code and production readiness
A guide for software engineers highlights the significant gap between AI-generated code that runs and code that is suitable for production. It emphasizes the importance of clear intent and specific prompts when using AI…
-
Anthropic SDK for TypeScript 迎来频繁更新,新增功能和错误修复
Anthropic 发布了其 TypeScript SDK 的多个更新,包括 v0.94.0 至 v0.90.0 版本。这些更新引入了工作负载身份联合、交互式 OAuth 等功能,并支持 claude-opus-4-7 等新模型。此次发布还包括对 Managed Agents API 的改进、通过环境变量设置标头的能力,以及 API 错误和 Bedrock 集成的错误修复。
-
Claude Code login fails on Windows due to OAuth timeout
Users are reporting an issue with Claude Code on Windows where the login process fails due to an OAuth timeout error. This problem prevents users from accessing the application, with the error message "OAuth error: time…
-
Cloudflare offers AI infrastructure with global network for LLM processing
Cloudflare has introduced new infrastructure aimed at supporting AI applications, particularly large language models. The system leverages Cloudflare Workers and Durable Objects to manage state and compute at the edge, …