### 亮点
- **GPT-5.1-Codex-Max**：推出我们最新的前沿智能编码模型。GPT-5.1-Codex-Max 提供了更高的可靠性、更快的迭代速度，并支持长期运行的项目级工作流。了解更多请访问 https://www.openai.com/index/gpt-5-1-codex-max
- **原生压缩支持**：新增对压缩的一流支持，提升了长时间编码会话的性能。
- **扩展的工具令牌限制**：Codex 模型现在支持最多 10,000 个工具输出令牌。可通过 `config.toml` 中的 `tool_output_token_limit` 进行配置。
- **Windows 代理模式**：为原生 Windows 引入了代理模式。Codex 可以在您的工作文件夹中读取文件、写入文件和运行命令，且需要更少的审批。代理模式使用新的实验性 Windows 沙盒来限制文件系统和网络访问。了解更多请访问 https://developers.openai.com/codex/windows
- **TUI 与用户体验改进**
* 消除了当没有 `git` 仓库时的幽灵快照通知。
* Codex Resume 现在会显示分支，并遵守当前工作目录进行筛选。
* 为图片添加了占位符图标。
* 现在可以在 `/status` 中查看积分。
### 已合并的 PR
- 修复：不在换行符处截断 (#6907) — @aibrahim
- 特性：荒野中的 arcticfox (#6906) — @aibrahim
- [应用服务器] 在 thread/resume 时填充 thread>turns>items (#6848) — @owenlin0
- 微调：无用的调试日志 (#6898) — @jif
- 修复(核心) 在对话前支持更改 /approvals (#6836) — @dylan.hurd
- 日常维护：整合压缩令牌使用情况 (#6894) — @jif
- 日常维护(应用服务器) 世界可写的 Windows 通知 (#6880) — @dylan.hurd
- 修复：并行工具调用指令注入 (#6893) — @jif
- 微调：稳定性 (#6895) — @jif
- 特性：大型提交警告 (#6838) — @jif
- 修复标签 (#6892) — @tibo
- 移动 Shell 以使用 `truncate_text` (#6842) — @aibrahim
- 运行远程自动压缩 (#6879) — @pakrym
- 不稳定的 unified_exec_formats_large_output_summary (#6884) — @aibrahim
- shell_command 返回自由格式输出 (#6860) — @pakrym
- 日常维护(核心) arcticfox (#6876) — @dylan.hurd
- 修复(tui) 幽灵快照通知 (#6881) — @dylan.hurd
- 修复：模型选择器中的拼写错误 (#6859) — @aibrahim
- 日常维护：更新 windows 文档 URL (#6877) — @ae
- 特性：调整 windows 沙盒字符串 (#6875) — @ae
- 修复：向 ThreadStartResponse 和 ThreadResumeResponse 添加更多字段 (#6847) — @mbolin
- 日常维护：更新 windows 沙盒文档 (#6872) — @ae
- 默认开启远程压缩 (#6866) — @pakrym
- [应用服务器] 引入 `turn/completed` v2 事件 (#6800) — @celia
- 更新积分状态详情 (#6862) — @zhao
- tui: 向 'codex resume' 添加分支，按 cwd 过滤 (#6232) — @172423086+nornagon-openai
- 浏览器漏洞冒烟测试，Windows 安全文档初稿 (#6822) — @iceweasel
- windows 沙盒：支持多个工作区根目录 (#6854) — @iceweasel
- 更新 codex 后端模型 (#6855) — @zhao
- exec-server (#6630) — @172423086+nornagon-openai
- 修复测试，使其不会在源代码树中生成多余的 `config.toml` (#6853) — @etraut
- [应用服务器-测试客户端] 特性：自动批准命令 (#6852) — @owenlin0
- 提升 `generated_ts_has_no_optional_nullable_fields` 测试的运行时间 (#6851) — @etraut
- 修复：本地压缩 (#6844) — @jif
- 修复 config.md 中关于 MCP 服务器的拼写错误 (#6845) — @simcoea
- [codex][otel] 支持 mtls 配置 (#6228) — @apanasenko
- 特性：在应用服务器中进行审查 (#6613) — @jif
- 日常维护(配置) 启用 shell_command (#6843) — @dylan.hurd
- 在选择自动模式时提示开启 windows 沙盒。 (#6618) — @iceweasel
- 添加按令牌截断的实用程序 (#6746) — @aibrahim
- 更新 faq.md 中关于支持模型的部分 (#6832) — @adpena
- 修复 localshell 工具调用 (#6823) — @zhao
- 特性：启用并行工具调用 (#6796) — @jif
- 特性：远程压缩 (#6795) — @jif
- 微调：应用服务器 (#6830) — @jif
- 微调：将幽灵提交标记为稳定 (#6833) — @jif
- 特性：git 分支工具 (#6831) — @jif
- :bug: 修复(rmcp-客户端)：使用 expires_at 刷新 OAuth 令牌 (#6574) — @LaelLuo
- 修复(windows) 在 Windows 上的 shell_command 及轻微解析问题 (#6811) — @dylan.hurd
- 日常维护(核心) 增加 shell_serialization 覆盖率 (#6810) — @dylan.hurd
- 将默认值更新为 gpt-5.1 (#6652) — @aibrahim
- 将函数调用负载日志降级为调试，以避免 noisy error-level stderr (#6808) — @cassirer
- execpolicy2 扩展 (#6627) — @zhao
- [应用服务器] 特性：添加 v2 命令执行审批流程 (#6758) — @owenlin0
- 后台速率限制获取 (#6789) — @zhao
- 将 cap_sid 文件移动到 ~/.codex 中，以便沙盒无法覆盖它 (#6798) — @iceweasel
- 修复 Windows 上 Alt-Gr 的 TUI 问题 (#6799) — @etraut
- core：添加禁用 shell 工具的功能 (#6481) — @172423086+nornagon-openai
- 日常维护(核心) 更新 shell 指令 (#6679) — @dylan.hurd
- 修复：用 camelCase 注释所有应用服务器 v2 类型 (#6791) — @owenlin0
- LM Studio OSS 支持 (#2312) — @rugved
- [应用服务器] 向 readme 添加事件 (#6690) — @celia
- core/tui: 非阻塞 MCP 启动 (#6334) — @172423086+nornagon-openai
- 日常维护：删除 chatwidget::tests::binary_size_transcript_snapshot tui 测试 (#6759) — @owenlin0
- 特性：execpolicy v2 (#6467) — @zhao
- 微调：个人 git 忽略 (#6787) — @jif
- 临时：为 windows 2 放弃 sccache (#6775) — @jif
- 特性：为无法解码的图片添加占位符以防止 400 错误 (#6773) — @jif
- 修复(核心) 序列化 shell_command (#6744) — @dylan.hurd
- 修复 FreeBSD/OpenBSD 构建：针对目标平台的 keyring 功能和 BSD 加固 (#6680) — @jinxiaoyong
- 免除 "codex" github 用户签署 CLA (#6724) — @etraut
- 日常维护(deps)：将 actions/github-script 从 7 升级到 8 (#6755) — @49699333+dependabot[bot]
- 修复：Claude 模型因处理空 finish_reason 而返回不完整响应 (#6728) — @dulikaifazr
- 修复 Windows Codex 终端上的 AltGr/反斜杠输入问题 (#6720) — @pornotato
- 还原 "临时：为 windows 放弃 sccache (#6673)" (#6751) — @etraut
- 修复：为基于脚本的工具解析 Windows MCP 服务器执行问题 (#3828) — @jlee14281
- 修复自定义提示命名参数的文档错误并添加规范示例 (#5910) — @169171880+Sayeem3051
- 收紧双重截断时的 panic (#6701) — @aibrahim
- 改进压缩功能 (#6692) — @aibrahim
- 将截断助手重构到其自己的文件中 (#6683) — @aibrahim
- 还原 "用于验证/提交 winget 包的模板和构建步骤" (#6696) — @aibrahim
- ci: 仅对 openai 组织仓库运行 CLA 助手 (#6687) — @joshka
- 处理入职流程中的 "不信任" 目录选择 (#4941) — @viniciusmotta8
- 忽略 unified_exec_respects_workdir_override (#6693) — @pakrym
- 在输入之前排序输出 (#6691) — @pakrym
- 特性：添加用于内部使用的 app-server-test-client crate (#5391) — @owenlin0
- 修复 codex 检测，添加新的以安全为重点的冒烟测试。 (#6682) — @iceweasel
- 特性(ts-sdk)：允许覆盖 CLI 环境 (#6648) — @lopopolo
- 用于验证/提交 winget 包的模板和构建步骤 (#6485) — @iceweasel
- 添加测试超时 (#6612) — @pakrym
- 默认启用 TUI 通知 (#6633) — @172423086+nornagon-openai
- 临时：为 windows 放弃 sccache (#6673) — @jif
- [应用服务器] 添加 mcp 工具调用项开始/完成事件 (#6642) — @celia
- 特性：缓存分词器 (#6609) — @jif
- 特性：为 unified_exec 提供更好的 UI (#6515) — @jif
- 特性：在执行 /new 时添加恢复日志 (#6660) — @jif
- 测试：替换 mount_sse_once_match 为 mount_sse_once 用于 SSE 模拟 (#6640) — @pakrym
- 为套件测试提升共享助手 (#6460) — @aibrahim
- 在聊天补全测试中使用共享网络门控助手 (#6461) — @aibrahim
- 避免双重截断 (#6631) — @aibrahim
- 还原 "还原 \"彻底检修 shell 检测并集中统一执行的命令生成\"" (#6607) — @pakrym
- [应用服务器] 对 JSON 模式导出和 one-of 类型的小修复 (#6614) — @owenlin0
- [应用服务器] 添加新的 v2 事件：`item/reasoning/delta`、`item/agentMessage/delta` 和 `item/reasoning/summaryPartAdded` (#6559) — @celia
- 日常维护(核心) 整合 apply_patch 测试 (#6545) — @dylan.hurd
- 仅列出失败的测试 (#6619) — @pakrym
- 特性：为 exec 和 TypeScript SDK 添加 --add-dir 支持 (#6565) — @33551757+danfhernandez
- 为 TypeScript SDK 添加 AbortSignal 支持 (#6378) — @33551757+danfhernandez
- 启用 close-stale-contributor-prs.yml 工作流 (#6615) — @pakrym
- 更新默认 yield 时间 (#6610) — @pakrym
- 关闭陈旧 PR 的工作流 (#6594) — @pakrym
- 将提示缓存测试迁移到 test_codex (#6605) — @pakrym
- 还原 "彻底检修 shell 检测并集中统一执行的命令生成" (#6606) — @pakrym
- 彻底检修 shell 检测并集中统一执行的命令生成 (#6577) — @pakrym
### Highlights
- GPT-5.1-Codex-Max: introducing our newest frontier agentic coding model. GPT-5.1-Codex-Max delivers higher reliability, faster iteration, and support for long-running, project-scale workflows. Learn more at https://www.openai.com/index/gpt-5-1-codex-max
- Native Compaction Support: added first-class support for Compaction, improving performance across extended coding sessions.
- Expanded Tool Token Limits: codex models now support up to 10,000 tool output tokens. Configure this via `tool_output_token_limit` in `config.toml`.
- Windows Agent mode: Introduced Agent mode for native Windows. Codex can read files, write files, and run commands in your working folder with fewer approvals. Agent mode uses a new experimental Windows sandbox to limit filesystem and network access. Learn more at https://developers.openai.com/codex/windows
- TUI and UX Improvements
* Eliminated ghost snapshot notifications when no `git` repository is present.
* Codex Resume now displays branches and respects the current working directory for filtering.
* Added placeholder icons for images.
* Credits are now visible in `/status`.
### PRs Merged
- fix: don't truncate at new lines (#6907) — @aibrahim
- feat: arcticfox in the wild (#6906) — @aibrahim
- [app-server] populate thread>turns>items on thread/resume (#6848) — @owenlin0
- nit: useless log to debug (#6898) — @jif
- fix(core) Support changing /approvals before conversation (#6836) — @dylan.hurd
- chore: consolidate compaction token usage (#6894) — @jif
- chore(app-server) world-writable windows notification (#6880) — @dylan.hurd
- fix: parallel tool call instruction injection (#6893) — @jif
- nit: stable (#6895) — @jif
- feat: warning large commits (#6838) — @jif
- fix label (#6892) — @tibo
- Move shell to use `truncate_text` (#6842) — @aibrahim
- Run remote auto compaction (#6879) — @pakrym
- flaky-unified_exec_formats_large_output_summary (#6884) — @aibrahim
- shell_command returns freeform output (#6860) — @pakrym
- chore(core) arcticfox (#6876) — @dylan.hurd
- fix(tui) ghost snapshot notifications (#6881) — @dylan.hurd
- fix: typos in model picker (#6859) — @aibrahim
- chore: update windows docs url (#6877) — @ae
- feat: tweak windows sandbox strings (#6875) — @ae
- fix: add more fields to ThreadStartResponse and ThreadResumeResponse (#6847) — @mbolin
- chore: update windows sandbox docs (#6872) — @ae
- Remote compaction on by-default (#6866) — @pakrym
- [app-server] introduce `turn/completed` v2 event (#6800) — @celia
- update credit status details (#6862) — @zhao
- tui: add branch to 'codex resume', filter by cwd (#6232) — @172423086+nornagon-openai
- smoketest for browser vuln, rough draft of Windows security doc (#6822) — @iceweasel
- windows sandbox: support multiple workspace roots (#6854) — @iceweasel
- updating codex backend models (#6855) — @zhao
- exec-server (#6630) — @172423086+nornagon-openai
- Fix tests so they don't emit an extraneous `config.toml` in the source tree (#6853) — @etraut
- [app-server-test-client] feat: auto approve command (#6852) — @owenlin0
- Improved runtime of `generated_ts_has_no_optional_nullable_fields` test (#6851) — @etraut
- fix: local compaction (#6844) — @jif
- Fix typo in config.md for MCP server (#6845) — @simcoea
- [codex][otel] support mtls configuration (#6228) — @apanasenko
- feat: review in app server (#6613) — @jif
- chore(config) enable shell_command (#6843) — @dylan.hurd
- Prompt to turn on windows sandbox when auto mode selected. (#6618) — @iceweasel
- Add the utility to truncate by tokens (#6746) — @aibrahim
- Update faq.md section on supported models (#6832) — @adpena
- fixing localshell tool calls (#6823) — @zhao
- feat: enable parallel tool calls (#6796) — @jif
- feat: remote compaction (#6795) — @jif
- nit: app server (#6830) — @jif
- nit: mark ghost commit as stable (#6833) — @jif
- feat: git branch tooling (#6831) — @jif
- :bug: fix(rmcp-client): refresh OAuth tokens using expires_at (#6574) — @LaelLuo
- fix(windows) shell_command on windows, minor parsing (#6811) — @dylan.hurd
- chore(core) Add shell_serialization coverage (#6810) — @dylan.hurd
- Update defaults to gpt-5.1 (#6652) — @aibrahim
- Demote function call payload log to debug to avoid noisy error-level stderr (#6808) — @cassirer
- execpolicy2 extension (#6627) — @zhao
- [app-server] feat: add v2 command execution approval flow (#6758) — @owenlin0
- background rate limits fetch (#6789) — @zhao
- move cap_sid file into ~/.codex so the sandbox cannot overwrite it (#6798) — @iceweasel
- Fix TUI issues with Alt-Gr on Windows (#6799) — @etraut
- core: add a feature to disable the shell tool (#6481) — @172423086+nornagon-openai
- chore(core) Update shell instructions (#6679) — @dylan.hurd
- fix: annotate all app server v2 types with camelCase (#6791) — @owenlin0
- LM Studio OSS Support (#2312) — @rugved
- [app-server] add events to readme (#6690) — @celia
- core/tui: non-blocking MCP startup (#6334) — @172423086+nornagon-openai
- chore: delete chatwidget::tests::binary_size_transcript_snapshot tui test (#6759) — @owenlin0
- feat: execpolicy v2 (#6467) — @zhao
- nit: personal git ignore (#6787) — @jif
- tmp: drop sccache for windows 2 (#6775) — @jif
- feat: placeholder for image that can't be decoded to prevent 400 (#6773) — @jif
- fix(core) serialize shell_command (#6744) — @dylan.hurd
- Fix FreeBSD/OpenBSD builds: target-specific keyring features and BSD hardening (#6680) — @jinxiaoyong
- Exempt the "codex" github user from signing the CLA (#6724) — @etraut
- chore(deps): bump actions/github-script from 7 to 8 (#6755) — @49699333+dependabot[bot]
- Fix: Claude models return incomplete responses due to empty finish_reason handling (#6728) — @dulikaifazr
- Fix AltGr/backslash input on Windows Codex terminal (#6720) — @pornotato
- Revert "tmp: drop sccache for windows (#6673)" (#6751) — @etraut
- fix: resolve Windows MCP server execution for script-based tools (#3828) — @jlee14281
- Fix documentation errors for Custom Prompts named arguments and add canonical examples (#5910) — @169171880+Sayeem3051
- Tighten panic on double truncation (#6701) — @aibrahim
- Improve compact (#6692) — @aibrahim
- Refactor truncation helpers into its own file (#6683) — @aibrahim
- Revert "templates and build step for validating/submitting winget package" (#6696) — @aibrahim
- ci: only run CLA assistant for openai org repos (#6687) — @joshka
- Handle "Don't Trust" directory selection in onboarding (#4941) — @viniciusmotta8
- Ignore unified_exec_respects_workdir_override (#6693) — @pakrym
- Order outputs before inputs (#6691) — @pakrym
- feat: add app-server-test-client crate for internal use (#5391) — @owenlin0
- fix codex detection, add new security-focused smoketests. (#6682) — @iceweasel
- feat(ts-sdk): allow overriding CLI environment (#6648) — @lopopolo
- templates and build step for validating/submitting winget package (#6485) — @iceweasel
- Add test timeout (#6612) — @pakrym
- Enable TUI notifications by default (#6633) — @172423086+nornagon-openai
- tmp: drop sccache for windows (#6673) — @jif
- [App server] add mcp tool call item started/completed events (#6642) — @celia
- feat: cache tokenizer (#6609) — @jif
- feat: better UI for unified_exec (#6515) — @jif
- feat: add resume logs when doing /new (#6660) — @jif
- tests: replace mount_sse_once_match with mount_sse_once for SSE mocking (#6640) — @pakrym
- Promote shared helpers for suite tests (#6460) — @aibrahim
- Use shared network gating helper in chat completion tests (#6461) — @aibrahim
- Avoid double truncation (#6631) — @aibrahim
- Revert "Revert \"Overhaul shell detection and centralize command generation for unified exec\"" (#6607) — @pakrym
- [app-server] small fixes for JSON schema export and one-of types (#6614) — @owenlin0
- [App-server] add new v2 events:`item/reasoning/delta`, `item/agentMessage/delta` & `item/reasoning/summaryPartAdded` (#6559) — @celia
- chore(core) Consolidate apply_patch tests (#6545) — @dylan.hurd
- Only list failed tests (#6619) — @pakrym
- feat: Add support for --add-dir to exec and TypeScript SDK (#6565) — @33551757+danfhernandez
- Add AbortSignal support to TypeScript SDK (#6378) — @33551757+danfhernandez
- Enable close-stale-contributor-prs.yml workflow (#6615) — @pakrym
- Update default yield time (#6610) — @pakrym
- Close stale PRs workflow (#6594) — @pakrym
- Migrate prompt caching tests to test_codex (#6605) — @pakrym
- Revert "Overhaul shell detection and centralize command generation for unified exec" (#6606) — @pakrym
- Overhaul shell detection and centralize command generation for unified exec (#6577) — @pakrym