agents.changelog

synced live · 92 entries

Official agent updates · explained

Latest official AI agent updates and release notes.

The authoritative changelog feed for 5 AI agent products, paired with plain-English context so you know what actually shipped.

Latest updates

Claude Code

Jul 10, 2026

New

Claude Code 2.1.206

Added directory path suggestions to /cd, matching /add-dir behavior

解读新增/cd命令的目录路径自动补全功能,现在输入/cd时会出现目录建议,类似于/add-dir的体验,让你能更快切换到目标目录。

OpenAI Codex

Jul 9, 2026

3 updates

Codex joins the ChatGPT desktop app

Codex is now part of the ChatGPT desktop app on macOS and Windows. Existing Codex app users can update as usual and keep their projects, settings, and workflows. You can make Codex the default view and, on macOS, keep the Codex app icon.

解读Codex 现在集成到了 ChatGPT 桌面应用中(支持 macOS 和 Windows)。原有的 Codex 用户更新后可以保留项目、设置和工作流程。你可以将 Codex 设为默认视图,在 macOS 上还能保留原来的图标。

Codex CLI 0.144.0

New Features: Usage-limit reset credits now show their type and expiration, and let you choose which credit to redeem.

解读现在可以看到重置积分的类型和有效期,并且可以自主选择要兑换哪个积分,方便管理。

Codex CLI 0.144.1

no notable user-facing improvements in this release

解读此版本无值得关注的用户可见改进

Claude Code

Jul 8, 2026

2 updates

Claude Code 2.1.204

no notable user-facing improvements in this release

解读此版本无值得关注的用户可见改进

Claude Code 2.1.205

Added an auto mode rule that blocks tampering with session transcript files

解读新增自动模式规则,防止会话记录文件被篡改,增强安全性。

Hermes Agent

Jul 8, 2026

2 updates

Hermes Agent v0.18.1 (2026.7.7)

no notable user-facing improvements in this release

解读此版本无值得关注的用户可见改进

Hermes Agent v0.18.2 (2026.7.7.2)

no notable user-facing improvements in this release

解读此版本无值得关注的用户可见改进

OpenAI Codex

Jul 8, 2026

Codex CLI 0.143.0

New Features: Remote plugins are now enabled by default, with richer catalog rows, npm marketplace sources, and visible remote/local versions.

解读远程插件现在默认开启,并且插件目录展示更丰富,支持从 npm 市场安装,还能显示远程和本地版本。这意味着你可以更方便地使用社区或自己发布的远程插件,同时能清晰看到版本差异。

Claude Code

Jul 7, 2026

Claude Code 2.1.203

Added a warning when your login is about to expire, so you can re-authenticate before background sessions are interrupted

解读新增登录即将过期的警告,方便用户在后台会话中断前重新认证,提升使用体验。

OpenAI Codex

Jul 6, 2026

ChatGPT for iOS

Added support for creating, searching, opening, forking, and managing Codex tasks directly from a conversation.

解读现在你可以直接在对话中创建、搜索、打开、复制和管理Codex任务,无需离开聊天界面,操作更快捷。

Claude Code

Jul 6, 2026

Claude Code 2.1.202

Added a "Dynamic workflow size" setting in /config for controlling how large Claude generally makes dynamic workflows (small/medium/large agent counts) — an advisory guideline, not an enforced cap

解读新增动态工作流大小设置,你可以在 /config 中选择小/中/大来控制工作流使用的代理数量,这是一个参考指南而非强制限制。

Claude Code

Jul 3, 2026

2 updates

Claude Code 2.1.200

Changed AskUserQuestion dialogs to no longer auto-continue by default; opt into an idle timeout via /config

解读当 Claude Code 需要向你提问时,默认不再自动继续执行。你可以通过 /config 命令设置空闲超时时间,让对话在指定时间内无操作后自动继续。这让你有更多时间思考问题,而不会被对话自动推进。

Claude Code 2.1.201

no notable user-facing improvements in this release

解读此版本无值得关注的用户可见改进

Claude Code

Jul 2, 2026

Claude Code 2.1.199

Stacked slash-skill invocations like /skill-a /skill-b do XYZ now load all leading skills (up to 5), not just the first

解读之前使用堆叠的斜杠技能调用(如 /skill-a /skill-b)时,只会加载第一个技能。现在会加载所有前导技能(最多5个),这意味着你可以一次性调用多个技能,提高效率。

Claude Code

Jul 1, 2026

Claude Code 2.1.198

Subagents now run in the background by default, so Claude keeps working while they run and is notified when they finish (previously a gradual rollout)

解读子代理现在默认在后台运行,这样Claude可以在它们运行的同时继续工作,并在完成后收到通知。这意味着你可以让Claude同时处理多个任务而不会阻塞主会话。

OpenAI Codex

Jul 1, 2026

Codex CLI 0.142.5

no notable user-facing improvements in this release

解读此版本无值得关注的用户可见改进

Hermes Agent

Jul 1, 2026

Hermes Agent v0.18.0 (2026.7.1) — The Judgment Release

✨ Highlights: Mixture-of-Agents is now a first-class model you can pick — MoA used to be a mode you toggled; now every named MoA preset shows up as a selectable model under a moa provider, right alongside Claude, GPT, and Grok in every model picker (CLI, TUI, desktop, gateway). Pick "my-council" the same way you'd pick any model, and Hermes routes your prompt through that ensemble automatically. An ensemble of frontier models deliberating on your hardest questions is now one selection away, on every surface. ( — @teknium1)

解读混合代理(MoA)现在成为可选模型:您可以在任何模型选择器中选择预设的 MoA 组合,Hermes 会自动将您的提示路由到该组合。这意味着您只需一次选择,即可让多个前沿模型共同推理您的问题。

Claude Code

Jun 30, 2026

Claude Code 2.1.197

Introducing Claude Sonnet 5: now the default model in Claude Code, with a native 1M-token context window and promotional pricing of $2/$10 per Mtok through August 31. Update to version 2.1.197 for access. https://www.anthropic.com/news/claude-sonnet-5

解读Claude Code 现在默认使用新推出的 Claude Sonnet 5 模型,该模型拥有原生 100 万 token 的上下文窗口(可以一次性处理超长文本),并享受促销定价:输入每 100 万 token 2 美元,输出每 100 万 token 10 美元,优惠持续到 8 月 31 日。升级到 2.1.197 版本即可体验。例如,你可以在一个对话中上传整本书或大型代码库,模型仍能准确理解上下文。

OpenClaw

Jun 30, 2026

OpenClaw 2026.6.11

Channel delivery reliability: Feishu voice replies from OpenClaw now show their duration in the chat bubble, so recipients can see how long the audio is before playing it. Related . Thanks @areslp, @fxz26284407, @kinrocw.

解读OpenClaw发送的飞书语音回复现在会显示持续时间,让接收者能提前知道语音长度,方便决定是否播放。

Claude Code

Jun 29, 2026

Claude Code 2.1.196

Added support for organization default models — admins set it in the org console; it shows as "Org default" (or "Role default") in /model when you haven't picked one yourself

解读新增了组织默认模型功能。管理员可以在组织控制台中设置默认模型,当你没有选择模型时,在 /model 中会显示为“组织默认”(或“角色默认”),简化了团队使用时的模型选择。

Claude Code

Jun 26, 2026

Claude Code 2.1.195

Added CLAUDECODEDISABLEMOUSECLICKS to disable mouse click/drag/hover in fullscreen mode while keeping wheel scroll

解读新增环境变量,可禁用全屏模式下的鼠标点击、拖拽、悬停,只保留滚轮。适合需要在全屏模式下防止误触的场景。

OpenAI Codex

Jun 25, 2026

3 updates

Codex Remote reaches general availability

Codex Remote has reached general availability. Use Codex from the ChatGPT mobile app to start or continue work on a connected Mac or Windows host, review progress, and approve actions from your phone.

解读Codex Remote 功能已正式发布。现在您可以通过 ChatGPT 手机应用连接已配对的 Mac 或 Windows 电脑,远程启动或继续工作、查看进展并在手机上批准操作。这极大提升了移动办公的灵活性,例如您可以在手机上让 Codex 在电脑端执行代码并查看结果。

Codex CLI 0.142.1

New Features: Added opt-in Windows system proxy support for authentication, including PAC, WPAD, static proxies, and bypass rules.

解读添加了可选的Windows系统代理支持,用于认证,包括PAC、WPAD、静态代理和绕过规则。这意味着你现在可以在Codex CLI中利用系统配置的代理(如公司网络代理)进行身份验证,而无需手动设置代理。例如,如果你的公司使用PAC文件自动配置代理,Codex CLI现在可以自动使用它。

Codex CLI 0.142.2

New Features: MCP tools now use tool search by default when supported, improving tool discovery while preserving compatibility with older models and providers.

解读当 MCP 工具支持时,现在默认启用工具搜索功能,让你能更方便地发现和调用可用工具,同时保持与旧模型和提供商的兼容。例如,你可以直接搜索“天气工具”而不需要手动查找。

Claude Code

Jun 25, 2026

Claude Code 2.1.193

Added autoMode.classifyAllShell setting to route all Bash/PowerShell commands through the auto-mode classifier instead of only arbitrary-code-execution patterns

解读新增了一个设置选项,让所有Bash/PowerShell命令都经过自动模式分类器处理,而不仅仅是那些可能执行任意代码的命令。这意味着更全面的命令管控,提高安全性。

Claude Code

Jun 24, 2026

2 updates

Claude Code 2.1.191

Added /rewind support for resuming a conversation from before /clear was run

解读新增了/rewind命令,允许你在执行/clear清除对话后,仍然能回到之前的对话状态。这意味着如果你不小心使用了/clear,可以用/rewind找回之前的聊天内容,避免信息丢失。

Claude Code 2.1.190

no notable user-facing improvements in this release

解读此版本无值得关注的用户可见改进

OpenClaw

Jun 24, 2026

OpenClaw 2026.6.10

Highlights: Automatic fast mode for talks: OpenClaw can enable fast mode for short conversational turns, then return to normal mode for longer runs with bounded fallback and delivery behavior. Thanks @alexph-dev and @vincentkoc.

解读新增自动快速模式,用于简短对话轮次,然后自动恢复正常模式。例如用户快速问答时,系统自动切换快速模式,提高响应效率。

Claude Code

Jun 23, 2026

Claude Code 2.1.187

Added sandbox.credentials setting to block sandboxed commands from reading credential files and secret environment variables

解读新增了 sandbox.credentials 设置,可以阻止沙箱命令读取凭据文件和秘密环境变量,提升安全性,避免敏感信息泄露。

Claude Code

Jun 22, 2026

Claude Code 2.1.186

Added claude mcp login and claude mcp logout to authenticate MCP servers from the CLI without opening the interactive /mcp menu, with --no-browser stdin redirect support for completing over SSH

解读新增 `claude mcp login` 和 `claude mcp logout` 命令,让你可以直接在命令行中登录或登出 MCP 服务器,不再需要打开交互式的 /mcp 菜单。此外,还支持 `--no-browser` 和标准输入重定向,方便在 SSH 会话中完成身份验证。这意味着在远程服务器或无图形界面的环境中操作更简单了。

OpenAI Codex

Jun 22, 2026

2 updates

ChatGPT for iOS

Added per-host personality settings with Friendly and Pragmatic options.

解读新增每个主机的个性设置,提供友好和务实两种选项,让AI的对话风格可以根据不同主机进行调整,更灵活。

Codex CLI 0.142.0

New Features: /usage can now show and redeem earned usage-limit reset credits, with confirmation, retry, and refreshed availability states.

解读现在你可以使用 `/usage` 命令查看并兑换获得的额度重置积分,操作会带有确认、重试和刷新后的可用状态提示,让你更好地管理使用额度。

OpenClaw

Jun 21, 2026

OpenClaw 2026.6.9

Highlights: Richer Telegram delivery: Telegram now sends rich HTML, preserves rich markdown and sticker paths, renders progress drafts and command output more faithfully, normalizes HTML tables safely, and keeps mentions and spooled handlers on the right delivery path. Thanks @obviyus, @vincentkoc, @goutamadwant, @kesslerio, @NianJiuZst, @SweetSophia, @Marvinthebored, @aaajiao, @zhangqueping, and @jairrab.

解读现在Telegram可以发送富文本消息,包括HTML、Markdown格式和贴纸路径等,消息渲染更准确。

Claude Code

Jun 20, 2026

Claude Code 2.1.185

The stream-stall hint now reads "Waiting for API response · will retry in …" instead of "No response from API · Retrying in …", and triggers after 20s of silence instead of 10s

解读这条更新改进了API请求超时时的提示信息:现在显示“正在等待API响应·将在...后重试”而不是“无API响应·正在重试...”,并且将等待时间从10秒增加到20秒。这意味着当API响应较慢时,用户会看到更准确的提示,并且有更长的等待时间,避免过早重试,提升使用体验。

Claude Code

Jun 19, 2026

Claude Code 2.1.183

Improved auto mode safety: destructive git commands (git reset --hard, git checkout -- ., git clean -fd, git stash drop) are now blocked when you didn't ask to discard local work, git commit --amend is blocked when the commit wasn't made by the agent this session, and terraform destroy/pulumi destroy/cdk destroy are blocked unless you asked for the specific stack

解读自动模式现在会阻止您未要求丢弃本地更改时执行危险的git命令(如git reset --hard、git checkout -- .等),以及阻止修改不是由当前会话代理创建的提交,并且阻止执行未明确指定的基础设施销毁命令(terraform/pulumi/cdk destroy)。这提高了自动模式的安全性,防止意外损坏代码。

Hermes Agent

Jun 19, 2026

Hermes Agent v0.17.0 (v2026.6.19)

✨ Highlights: Hermes reaches iMessage — Photon Spectrum, no Mac relay required — There's now an iMessage platform plugin built on Photon's managed line pool. Run hermes photon login, authenticate with a device code, and Hermes can send and receive iMessage — no Mac sitting in a closet running a relay, no BlueBubbles bridge to babysit. It's positioned as the successor to BlueBubbles: free to start, nothing to self-host. If your friends and family live in the blue bubbles, Hermes lives there now too. ( — @teknium1)

解读Hermes 现在可以直接与 iMessage 通信,无需依赖 Mac 中继或第三方桥接。你只需通过 `hermes photon login` 进行设备码认证,即可发送和接收 iMessage,就像使用普通聊天工具一样。这意味着即使你的朋友和家人使用蓝色气泡(iMessage),你也能无缝互动。

OpenAI Codex

Jun 18, 2026

2 updates

Codex app

Added Record & Replay, a macOS feature that turns a demonstrated workflow into a reusable skill. Initial availability excludes the European Economic Area, the United Kingdom, and Switzerland. You or your administrator must also enable Computer Use.

解读新增了“记录与重放”功能,这是一种 macOS 特性,可以将你演示的工作流程转化为可重复使用的技能。首次发布时不包括欧洲经济区、英国和瑞士,并且需要你或管理员启用“计算机使用”权限。这意味着你可以录制操作步骤,然后让 Codex 自动执行这些步骤,提高效率。

Codex CLI 0.141.0

New Features: Remote executors now use authenticated, end-to-end encrypted Noise relay channels.

解读远程执行器现在使用经过认证的端到端加密的Noise中继通道,确保数据传输更安全。

Claude Code

Jun 17, 2026

Claude Code 2.1.181

Added /config key=value syntax to set any setting from the prompt (e.g. /config thinking=false) — works in interactive, -p, and Remote Control

解读新增了在提示中使用 `/config key=value` 语法直接修改设置的功能,例如 `/config thinking=false`,适用于交互式、-p 和远程控制模式,让用户无需离开对话即可调整配置。

OpenClaw

Jun 16, 2026

OpenClaw 2026.6.8

Highlights: Richer channel delivery: Telegram and WhatsApp are less brittle: Telegram renders structured text with tables, lists, expandable blockquotes, preserved intentional line breaks, and CLI-backed replies, while WhatsApp now honors configured ACP bindings. Thanks @obviyus, @vincentkoc, @jzakirov, @spacegeologist, @TurboTheTurtle, @mcaxtr, @myrzka, and @dmorn.

解读为Telegram消息传递增加了丰富的格式支持,例如表格、列表、可展开的引用块、保留换行以及CLI回复;同时让WhatsApp能够正确使用配置的ACP绑定,使消息更美观可靠。

OpenAI Codex

Jun 16, 2026

Codex app features are available in the EEA, UK, and Switzerland

More Codex app capabilities are rolling out to users in the European Economic Area, the United Kingdom, and Switzerland:

解读Codex 应用现在向欧洲经济区、英国和瑞士的用户推出更多功能,这些用户将能体验到之前未提供的增强能力。

Claude Code

Jun 16, 2026

Claude Code 2.1.179

no notable user-facing improvements in this release

解读此版本无值得关注的用户可见改进

OpenAI Codex

Jun 15, 2026

2 updates

ChatGPT for iOS

Added a workspace file browser for previewing files and linking workspace paths into prompts.

解读新增工作区文件浏览器,可预览文件并链接到提示中,方便快速引用工作区内容。

Codex CLI 0.140.0

New Features: Added /usage views for daily, weekly, and cumulative account token activity.

解读新增了 /usage 命令,可以查看每日、每周和累计的账户 token 使用情况,方便用户监控消耗。

Claude Code

Jun 15, 2026

Claude Code 2.1.178

Added Tool(param:value) syntax for permission rules to match a tool's input parameters (with * wildcard), e.g. Agent(model:opus) to block Opus subagents

解读新增了在权限规则中使用 Tool(param:value) 语法来匹配工具的输入参数,支持通配符 *,例如可以用 Agent(model:opus) 来阻止 Opus 子代理,让权限控制更精细。

Claude Code

Jun 13, 2026

Claude Code 2.1.177

no notable user-facing improvements in this release

解读此版本无值得关注的用户可见改进

Claude Code

Jun 12, 2026

3 updates

Claude Code 2.1.175

Added enforceAvailableModels managed setting — when enabled, the availableModels allowlist also constrains the Default model (a Default that would resolve to a disallowed model now falls back to the first allowed model), and user or project settings can no longer widen a managed availableModels list

解读新增了 `enforceAvailableModels` 管理设置。开启后,`availableModels` 白名单也会限制默认模型(如果默认模型不在白名单中,会自动切换为第一个允许的模型),同时用户或项目设置无法再扩大管理员设置的 `availableModels` 列表。这意味着管理员可以更严格地控制可用的模型,防止用户绕过限制。例如,管理员只允许 GPT-4 和 Claude-3,用户就无法通过项目设置添加 Llama。

Claude Code 2.1.174

Added wheelScrollAccelerationEnabled setting to disable mouse-wheel scroll acceleration in fullscreen mode

解读新增了一个设置 wheelScrollAccelerationEnabled,可以禁用全屏模式下的鼠标滚轮加速,让用户在全屏时滚动更符合自己的习惯。

Claude Code 2.1.176

Session titles are now generated in the language of your conversation (set the language setting to pin a specific language)

解读现在会话标题会根据对话使用的语言自动生成,你可以通过设置 `language` 来固定使用某种语言。这意味着标题更贴合对话内容,方便识别。

OpenClaw

Jun 12, 2026

OpenClaw 2026.6.6

Highlights: Telegram delivery is safer and more coherent: account-scoped topics route to the right agent, streamed text survives tool calls, /compact works on generic ingress, callback handling uses concrete APIs, draft chunking is shared, durable dispatch dedupe moved into the SDK, and unauthorized DM text stays out of cache and prompt context. Thanks @codysai001, @alexzhu0, @joelnishanth, @snowzlm, @obviyus, and @sallyom.

解读改进了Telegram消息传递功能:现在账号级话题能正确路由到对应的智能体,流式文本在工具调用后不会丢失,/compact命令在通用入口可用,回调处理使用具体API,草稿分块共享,持久分发去重移到SDK中,未经授权的私信文本不会进入缓存和提示上下文。

Claude Code

Jun 11, 2026

2 updates

Claude Code 2.1.173

no notable user-facing improvements in this release

解读此版本无值得关注的用户可见改进

Claude Code 2.1.175

Added `enforceAvailableModels` managed setting — when enabled, the `availableModels` allowlist also constrains the Default model (a Default that would resolve to a disallowed model now falls back to the first allowed model), and user or project settings can no longer widen a managed `availableModels` list

解读新增了 `enforceAvailableModels` 托管设置。启用后,`availableModels` 白名单也会限制默认模型:如果默认模型指向一个不在白名单中的模型,则会自动切换到白名单中的第一个可用模型。此外,用户或项目的本地设置不能扩大由管理员托管的 `availableModels` 列表。这增强了管理员对可用模型的控制,防止用户绕过限制。例如,管理员只允许使用模型 A 和 B,则用户无法通过自己的设置使用模型 C。

OpenAI Codex

Jun 11, 2026

Codex app

Added rate-limit reset banking for Plus and Pro users, including one free reset at launch and referral invitations for earning more during the current promotion. Eligible Business members can invite coworkers to earn shared workspace credits through a separate referral program.

解读新增速率限制重置银行功能,Plus和Pro用户可获得一次免费重置,并可通过邀请赚取更多重置次数。这意味着用户在高频使用时可以更灵活地应对限制。

Claude Code

Jun 10, 2026

Claude Code 2.1.172

Sub-agents can now spawn their own sub-agents (up to 5 levels deep)

解读子代理现在可以生成自己的子代理,最多可达5层深度。这意味着你可以构建更复杂的任务分解层次,让子代理再派生出更小的子代理来处理子任务。