Claude Code v2.1.250更新：恢复会话、受限模式与反馈报告
Week 35 · August 24–28, 2026
Claude Code持续迭代，这次更新的受限模式和会话恢复对工程化部署和混合工作流很有价值，开发者值得跟进。
Resume terminal sessions in the Claude Code Desktop app, review feedback reports that Claude drafts for you, and start a session in restricted mode.
在 Claude Code 桌面应用中恢复终端会话，查看 Claude 为你起草的反馈报告，并以受限模式启动会话。
Releases v2.1.240 → v2.1.250 3 features · August 24–28
发布 v2.1.240 → v2.1.250 3 个新功能 · 8月24日–28日
Resume terminal sessions in the Desktop app Desktop
在桌面应用中恢复终端会话 桌面端
Type /resume in the Claude Code Desktop prompt box to pick up any session you started from the CLI and continue it in the app with the full conversation and context intact. Search your sessions by title, folder, or branch, and preview where you left off before you resume.
在 Claude Code 桌面应用的提示框中输入 /resume，以拾取你从 CLI 启动的任何会话，并在应用中继续该会话，同时保留完整的对话和上下文。你可以按标题、文件夹或分支搜索会话，并在恢复之前预览你上次离开的地方。
In a Desktop session, run the command to list your terminal sessions:
在桌面会话中，运行以下命令列出你的终端会话：
```text Claude Code theme={null} > /resume ``````text Claude Code theme={null} > /resume ```Select a session and press Enter. The conversation opens in the app where you left off.
选择一个会话并按 Enter。对话将在你上次离开的地方于应用中打开。
Move between the CLI and Desktop
在 CLI 和桌面应用之间切换
Claude-drafted feedback CLI
Claude 起草的反馈 CLI
When a tool keeps failing, Claude can't help with a request, or you point out a mistake, Claude now drafts a feedback report for you with the SendFeedback tool. A card above your prompt shows the draft, and you can review, send, or dismiss it from there. Nothing reaches Anthropic until you send it. Requires v2.1.238 or later.
当工具持续失败、Claude 无法协助某个请求，或者你指出错误时，Claude 现在会使用 SendFeedback 工具为你起草一份反馈报告。提示框上方会显示草稿卡片，你可以在此处查看、发送或忽略它。在你发送之前，没有任何内容会发送给 Anthropic。需要 v2.1.238 或更高版本。
Run /feedback with no argument to open your queue of drafts from every session:
运行不带参数的 /feedback 以打开来自所有会话的草稿队列：
```text Claude Code theme={null} > /feedback ``````text Claude Code theme={null} > /feedback ```Select a draft, then edit, send, or discard it. To turn drafting off, set Claude-drafted feedback to off in /config.
选择一个草稿，然后编辑、发送或删除它。要关闭草稿功能，请在 /config 中将 Claude-drafted feedback 设置为 off。
SendFeedback tool behavior
SendFeedback 工具行为
Restricted mode v2.1.248
受限模式 v2.1.248
Restricted mode starts Claude Code without the built-in tools that run commands or code. Use it when an evaluation harness drives claude on a shared machine. Start it with `--restricted` or set CLAUDE\_CODE\_RESTRICTED=1. Claude Code also removes WebFetch, confines the file tools to the working directories, loads only managed settings and `--settings`, and refuses the bypassPermissions permission mode.
受限模式在不运行命令或代码的内置工具的情况下启动 Claude Code。当评估框架在共享机器上驱动 claude 时使用此模式。通过 `--restricted` 启动，或设置 CLAUDE_CODE_RESTRICTED=1。Claude Code 还会移除 WebFetch，将文件工具限制在工作目录内，仅加载受管设置和 `--settings`，并拒绝 bypassPermissions 权限模式。
Run a non-interactive query without the command-running tools:
在不使用命令运行工具的情况下运行非交互式查询：
```bash terminal theme={null} claude --restricted -p "review src/ for SQL injection risks" ``````bash terminal theme={null} claude --restricted -p "review src/ for SQL injection risks" ```To give Claude one of the removed tools back, list it in `--tools` together with the other built-in tools you want, for example `--tools "Bash,Read,Edit"`. `--tools` is an allowlist, and its default preset doesn't restore the removed tools.
要将其中一个被移除的工具重新赋予 Claude，请将其与你想使用的其他内置工具一起列在 `--tools` 中，例如 `--tools "Bash,Read,Edit"`。`--tools` 是一个白名单，其默认预设不会恢复被移除的工具。
CLI flags
CLI 标志
Other wins
其他改进
Set the new modelPicker setting to extend or replace the /model picker's built-in list with your own ordered, labeled entries, including Amazon Bedrock or Google Cloud's Agent Platform model IDs
将新的 modelPicker 设置设置为用你自己的有序、带标签的条目扩展或替换 /model 选择器的内置列表，包括 Amazon Bedrock 或 Google Cloud 的 Agent Platform 模型 ID
Set promptCacheTtl to 1h to keep a one-hour prompt cache on the main conversation when you use an API key or a cloud provider; subagentPromptCacheTtl sets the TTL for subagents and all other requests outside the main conversation
将 promptCacheTtl 设置为 1h，以便在使用 API 密钥或云提供商时在主对话中保留一小时提示缓存；subagentPromptCacheTtl 设置子代理以及主对话外所有请求的 TTL
On Pro, Max, Team, and Enterprise plans, /usage adds a Loops breakdown: run count, total tokens, tokens per run, and last run for the /loop and scheduled tasks that used the most tokens
在 Pro、Max、Team 和 Enterprise 套餐中，/usage 增加了 Loops 细分数据：运行次数、总 token 数、每次运行的 token 数，以及使用最多 token 的 /loop 和定时任务的最后一次运行时间
Organizations on contracted rates can set the modelPricing managed setting so /usage, the status line, and OpenTelemetry report cost at those rates instead of list price
采用合同费率的组织可以设置 modelPricing 托管配置项，使 /usage、状态行和 OpenTelemetry 报告按该费率而非目录价计算成本
/login offers Sign in with your Console account under the Anthropic Console account option, so members of Console organizations that don't allow API keys can sign in without creating one
/login 在 Anthropic Console 账户选项下提供使用你的 Console 账户登录的功能，因此不允许使用 API 密钥的 Console 组织成员可以在不创建密钥的情况下登录
Run /permissions and open the new Auto mode tab to view and edit auto mode classifier rules without opening a settings file
运行 /permissions 并打开新的 Auto 模式标签页，即可在不打开配置文件的情况下查看和编辑自动模式分类器规则
When auto mode is available, Bash permission prompts in the Manual and acceptEdits permission modes offer a Yes, and switch to auto mode option; select it to approve the command and switch the session to auto mode
当自动模式可用时，Manual 和 acceptEdits 权限模式下的 Bash 权限提示会提供“是，并切换到自动模式”选项；选择该选项可批准命令并将会话切换至自动模式
After you move a session with /cd, the new directory's project settings, hooks, .mcp.json servers, skills, and subagents take effect immediately instead of on the next `--resume`
通过 /cd 移动会话后，新目录的项目设置、hooks、.mcp.json 服务器、技能和子智能体会立即生效，而无需等到下次 `--resume`
In non-interactive sessions, including -p runs, Agent SDK runs, and cloud sessions, Claude Code continues a response that a server error, dropped connection, or stall cut off mid-stream, when the partial response contains text and no tool calls
在非交互式会话中（包括 -p 运行、Agent SDK 运行和云会话），如果部分响应包含文本且无工具调用，Claude Code 会继续由服务器错误、连接中断或停滞导致中途截断的响应
A subagent that stops at its maxTurns limit returns its output marked as partial, with a hint that Claude can continue it with SendMessage, instead of appearing finished
达到 maxTurns 限制而停止的子智能体会返回标记为 partial 的输出，并提示 Claude 可以通过 SendMessage 继续处理，而不是显示为已完成
On Amazon Bedrock, Google Cloud's Agent Platform, and Microsoft Foundry, sessions on the same machine can now message each other, /loop can choose its own interval, and /model and /effort apply immediately instead of after the turn ends
在 Amazon Bedrock、Google Cloud 的 Agent Platform 和 Microsoft Foundry 上，同一机器上的会话现在可以互相发送消息，/loop 可以自行选择间隔时间，/model 和 /effort 会立即生效，而无需等待回合结束
The native installer and auto-updater download a zstd-compressed build, about 75 MB instead of 340 MB on Linux x64, and native builds load code on demand, using roughly 40 to 70 MB less memory per session
原生安装程序和自动更新程序下载的是经过 zstd 压缩的构建版本，在 Linux x64 平台上大小约为 75 MB（而非 340 MB），且原生构建按需加载代码，每个会话内存占用减少约 40 至 70 MB
Full changelog for v2.1.240–v2.1.250 →
v2.1.240–v2.1.250 完整更新日志 →
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力