One read of Claude Documentationclaude-docs-20261015T153707Z
5 pages moved out of 266 read.
Pages moved
5
significant first
Pages read
266
in this capture
Captured
15:37 UTC
Corpus hash
a9500089be82
corpus-hash
What this read moved
1-5 of 5cowork/changelog Changed · +44 / -0 lines
from line 2
22
33> Release notes for Claude Desktop
44
5<Update label="v2.42304.0" description="2026-10-15">
6 Bundled Claude Code version: 2.1.299.
7
8 **General**
9
10 * Changed the certificate that signs the bundled Claude Code helper (`claude.exe`) on Windows to the renewed one that the app and its installer have used since v2.37337.1. Rules that hold the certificate itself, such as a Microsoft Defender certificate indicator, need the new certificate added beside the old one.
11 * Changed the import from Claude (third-party) where device management turned it on: after the import, Claude now adds local MCP servers and starts those that were on, installs the desktop extensions and uploaded plugins that your organization allows, and, where your organization allows skill creation, uploads skills that grant tools or declare hooks, all without asking you to confirm. Before, each local MCP server waited for you to confirm its command and settings. Organizations under the HIPAA restriction get none of these.
12 * Improved the import from Claude (third-party): copied scheduled tasks keep their attached files, and your Instructions for Claude and Cowork global instructions can be added from Settings > Import. Where device management turned it on, copied scheduled tasks also keep their folders and, while a local MCP server is unchanged, the approvals you gave for its tools.
13 * Fixed Arabic and Hebrew text reading in the wrong order in chat titles, session titles, messages, plans, search results, file views, feedback dialogs and the Code tab.
14 * Fixed some plugins never installing after being enabled on claude.ai because one of their files is very repetitive, such as a long empty table.
15 * Fixed some Update buttons planning a restart with nothing on screen while sessions were still running; they now ask first and can be cancelled.
16 * Fixed the app sometimes starting with no window open, including after restarting from the Help menu on Windows and Linux when the app had started automatically at sign-in.
17 * Fixed tool approval prompts staying hidden behind a full-screen MCP app after the chat preview was closed.
18
19 **Code**
20
21 * Changed the Git operations the app runs for you (the commit and stash around a branch switch, the in-folder branch switch, and the push behind "Continue in cloud") to no longer run repository Git hooks; push Git LFS objects from a terminal before continuing in the cloud.
22 * Removed the "Get started with Claude Code" checklist from the sidebar.
23 * Fixed "Rewind to here" failing with "Can’t rewind to this message." on any message but the latest in cloud and self-hosted sessions.
24 * Fixed deleting a session on an SSH host whose helper had not yet updated also deleting a branch holding unpushed commits.
25 * Fixed SSH sessions silently losing part of Claude's output, getting stuck reconnecting after another session on the same host failed to start, and ending the terminal on Windows computers when the host closed the connection.
26 * Fixed untracked files in a session's worktrees being deleted when the session was archived after Claude moved it to the cloud.
27
28 **Cowork**
29
30 * Changed computer use on macOS to no longer save files into hidden or protected folders, such as Homebrew's folders and shell start-up locations, or give a document a hidden or otherwise protected name.
31 * Changed Settings › Import to no longer offer the third-party app’s local MCP servers for organizations under the HIPAA restriction.
32 * Fixed computer use typing text twice into a busy web page in a background app.
33 * Fixed Enter selecting a slash menu item instead of sending the message when a "/" word, such as "/s", came after other words.
34 * Fixed the conversation sometimes jumping back to the bottom when you scroll up with Page Up, Shift+Space or the Up arrow key while Claude is responding.
35 * Fixed the workspace sometimes failing to start on Windows with a connection timeout error.
36
37 **3P**
38
39 * Added support in the macOS and Linux apps for SSH sessions carrying on across an app restart, and automatic recovery on every platform when Claude Code on the host is lost mid-task: the app restarts it and continues the task, even with nobody at the computer.
40 * Changed `claudeAiImport.automatic3pImport` to reinstall, without asking, the optional plugins that a user had switched on from any optional `allowedPluginMarketplaces` entry (GitHub, git, or a URL on any host) with the same source in the old and new organization, not only from the gateway or bootstrap origin. This includes computers that moved organizations before this version, at their first launch with it.
41 * Changed `mcpPersistentAlwaysAllowEnabled` to also apply to Code sessions on the device, in WSL and over SSH: when it is `false`, MCP tool prompts offer "Allow for this session" instead of "Always allow", and scheduled tasks offer no lasting approval unless `mcpScheduledTaskApprovalLifetimeDays` is one day or more.
42 * Changed Chat's file-analysis sandbox (`chatAdvancedFileAnalysisEnabled`) to reach the hosts in `coworkEgressAllowedHosts`, for example to install packages. As in Cowork, its commands run without an approval prompt. Before, it had no network access by default. With no list, or an empty one, it still has none.
43 * Changed session export in Settings > Import & export to cover everything by default instead of the last 30 days and to list what the file includes and leaves out; the import of a Claude Desktop export file also starts on everything.
44 * Fixed a message typed at the token limit being discarded in Cowork and Chat. The message shown at the limit now also says when the limit resets and, in Cowork and Chat, has a View usage button.
45 * Fixed Code tab SSH sessions on Windows hosts without Git for Windows refusing every shell command with "Enterprise policy requires sandboxing". The refusal stays for organizations that deny or ask on Bash commands by pattern, or whose `disabledBuiltinTools` holds `Bash` (with or without a pattern), `PowerShell` with a pattern, or a name the app does not recognize.
46 * Fixed connectors losing their saved sign-in when Claude starts while the macOS keychain is locked; the connector now says what to do.
47</Update>
48
549<Update label="v2.37337.1" description="2026-10-13">
650 Bundled Claude Code version: 2.1.295.
751
government/config/settings Changed · +10 / -4 lines
from line 132
132132
133133### Product availability
134134
135A group of separate switches that control which Claude products and features are available to members. Each switch appears as its own row: **Claude Desktop**, **Chat in Claude Desktop**, **Advanced file analysis in Chat**, **Cowork in Claude Desktop**, **Code in Claude Desktop**, **Claude Code**, and **Claude for Microsoft 365**. These switches are on by default, except for **Claude Code** and **Claude for Microsoft 365**.
135A group of separate switches that control which Claude products and features are available to members. Each switch appears as its own row: **Claude Desktop**, **Chat in Claude Desktop**, **Advanced file analysis in Chat**, **Cowork in Claude Desktop**, **Code in Claude Desktop**, **Scheduled tasks**, **Claude Code**, and **Claude for Microsoft 365**. These switches are on by default, except for **Claude Code** and **Claude for Microsoft 365**.
136136
137Turning off **Claude Code** or **Claude for Microsoft 365** stops the members the setting applies to from using that application. Turning off **Claude Desktop** stops the app from receiving your organization's settings. These switches turn applications off, not accounts, and are not a security boundary: to cut one member off from everything at once, deactivate their account (see [Deactivated users](/docs/government/org-admin/users#deactivated-users)). Turning off one of the other four switches removes that feature from Claude Desktop, as described below.
137Turning off **Claude Code** or **Claude for Microsoft 365** stops the members the setting applies to from using that application. Turning off **Claude Desktop** stops the app from receiving your organization's settings. These switches turn applications off, not accounts, and are not a security boundary: to cut one member off from everything at once, deactivate their account (see [Deactivated users](/docs/government/org-admin/users#deactivated-users)). Turning off any of the other switches removes that feature from Claude Desktop, as described below.
138138
139139The **Chat in Claude Desktop**, **Cowork in Claude Desktop**, and **Code in Claude Desktop** switches each make one part of the app available to members. Chat is for simple conversations, Cowork is for longer tasks that Claude works through on its own in a local workspace folder, and Code is for software development. The **Claude Code** switch is separate and applies to the standalone Claude Code command-line tool. To set up Claude Code on agency devices, see [Connect Claude Code to Claude for Government](/docs/government/deploy-claude-code/configure).
140140
from line 152
152152 **Claude Desktop home** needs Claude Desktop 1.52386.0 or later. Earlier versions ignore it and follow the switches.
153153</Note>
154154
155The **Scheduled tasks** switch controls whether members can set up tasks that Claude runs on a schedule. With it off, Claude Desktop hides the pages for them, Claude cannot schedule anything in the app, and tasks members already set up are kept but do not run.
156
157<Note>
158 **Scheduled tasks** needs Claude Desktop 2.110.0 or later. Earlier versions ignore it.
159</Note>
160
155161### Claude Code
156162
157163The **Claude Code** section of the **Config** page holds your organization's settings for Claude Code. Claude for Government delivers them to the command-line tool after a member signs in, and in Claude Desktop 2.9939.2 or later they can also reach Code sessions. While Claude Code is in [early access](/docs/government/deploy-claude-code/configure), the section appears once your agency has access.
from line 172
166172* **Opening from the browser**: whether links in web pages and editors can open Claude Code with a prompt filled in. **Blocked** by default.
167173* **Shell command sandbox**: whether members' shell commands run inside Claude Code's sandbox. The default, **Claude Code default**, leaves that to Claude Code's own settings. **Off** runs the commands without the sandbox, for workspaces where it cannot start, such as containers.
168174
169Related controls sit outside this section. Under [Tool and connector cards](#tool-and-connector-cards), the **Shell commands**, **Web fetch**, and **Web search** cards turn a whole tool off for Claude Code or make it ask every time. The **Member-added connectors** setting on the **Connectors** card sets which other connectors members can use in Claude Code, beyond web search and the ones you add there: **None** (the default), **Only these**, or **Any connector**. In Claude Desktop 2.2553.0 or later it also governs connectors delivered inside plugins: a remote one connects under **Any connector** or when its address is listed under **Only these**, and a local MCP server does not start under **None** or **Only these**. [Bypass-permissions mode and Auto mode](#bypass-permissions-mode-and-auto-mode) apply to Claude Code as described there.
175Related controls sit outside this section. Under [Tool and connector cards](#tool-and-connector-cards), the **Shell commands**, **Web fetch**, and **Web search** cards turn a whole tool off for Claude Code or make it ask every time. The **Member-added connectors** setting on the **Connectors** card sets which other connectors members can use in Claude Code, beyond web search and the ones you add there: **None** (the default), **Only these**, or **Any connector**. In Claude Desktop 2.2553.0 or later it also governs connectors delivered inside plugins: a remote one connects under **Any connector** or when its address is listed under **Only these**, and a local MCP server does not start under **None** or **Only these**. [Bypass-permissions mode and Auto mode](#bypass-permissions-mode-and-auto-mode) apply to Claude Code as described there. Under [Product availability](#product-availability), **Scheduled tasks** also turns off scheduling in Claude Code, such as `/loop`.
170176
171177Connectors you add on the **Connectors** card reach the Claude Code command-line tool when you tick **Claude Code** under **Apply to**, as [Connectors](/docs/government/connectors/overview) describes.
172178
third-party/claude-desktop/configuration Changed · +27 / -13 lines
from line 508
508508
509509<AccordionGroup>
510510 <Accordion title="chatAdvancedFileAnalysisEnabled details">
511 Also enables inline data analysis. Of the user's own files, the sandbox can read only those attached to the conversation, plus folders added to the chat's project through the app (read-only). By default it has no network access. Project folders are not available in the sandbox of a chat started while a rule on `Read`, `Grep` or `Glob` is set: in `disabledBuiltinTools`, in `builtinToolPolicy` with a value other than `allow`, or as a deny or ask rule in Claude Code's own managed settings on the device, whether or not the rule covers the folder. They are not available either where those managed settings cannot be read, or where Claude Code takes its managed settings from a gateway. Claude's file tools still read what the rules allow.
511 Also enables inline data analysis. Of the user's own files, the sandbox can read only those attached to the conversation, plus folders added to the chat's project through the app (read-only).
512
513 Commands in the sandbox can reach the hosts in `coworkEgressAllowedHosts`, and no others. If that list is unset or empty, they have no network access. With `*` in the list they can reach any host, as in Cowork. Plugin CLIs are the exception: they can always reach the hosts their plugin declares, plus the listed hosts (with `*`, only the declared hosts). Older app versions give the sandbox no network access by default.
514
515 As in Cowork, commands run without an approval prompt. So they can upload anything the sandbox can read to a listed host, and run what they download from one. To have Claude ask before each command, set `Bash` to `ask` in `builtinToolPolicy`, which also applies to Cowork and Code. To let commands install packages, list the hosts of the package manager's default index (for pip, `pypi.org` and `files.pythonhosted.org`). For a private mirror, list its host instead. Nothing points package managers at a mirror, so each command must name it (for pip, `--index-url`).
516
517 Project folders are not available in the sandbox of a chat started while a rule on `Read`, `Grep` or `Glob` is set: in `disabledBuiltinTools`, in `builtinToolPolicy` with a value other than `allow`, or as a deny or ask rule in Claude Code's own managed settings on the device, whether or not the rule covers the folder. They are not available either where those managed settings cannot be read, or where Claude Code takes its managed settings from a gateway. Claude's file tools still read what the rules allow.
512518 </Accordion>
513519
514520 <Accordion title="desktopHome details">
from line 543
537543 </Accordion>
538544
539545 <Accordion title="sshClientPath details">
540 Pins which OpenSSH client the app runs wherever it starts `ssh`: evaluating the user's SSH configuration (`ssh -G`), making the SSH connection and its channels, and the Code tab's terminal. `ssh-keygen` and `ssh-add` are taken from the same directory when they exist there, otherwise from PATH. The program must be OpenSSH 7.6 or newer (on Windows, Win32-OpenSSH 9.4 or newer to carry the connection); on macOS and Linux a wrapper script that ends in one is accepted, on Windows it must be a native `.exe` (not a .cmd, .bat or .ps1 script). When this key is set and the program is missing, cannot be run, or is too old, SSH sessions fail with an error telling the user to ask their IT administrator (the configured path is in its details) — the app never falls back to another ssh. The connection itself runs through this program on the system-OpenSSH engine, which the SSH connection engine setting's `system-openssh` value selects on every platform, including Windows; when the app's built-in SSH library makes the connection instead, this key still governs configuration evaluation (`ssh -G`), host-key lookups (`ssh-keygen`) and the terminal.
546 Pins which OpenSSH client the app runs wherever it starts `ssh`: evaluating the user's SSH configuration (`ssh -G`), making the SSH connection and its channels, and the Code tab's terminal. `ssh-keygen` and `ssh-add` are taken from the same directory when they exist there, otherwise from PATH. The program must be OpenSSH 7.6 or newer (on Windows, Win32-OpenSSH 9.4 or newer to carry the connection); on macOS and Linux a wrapper script that ends in one is accepted, on Windows it must be a native `.exe` (not a .cmd, .bat or .ps1 script). When this key is set and the program is missing, cannot be run, or is too old, SSH sessions fail with an error telling the user to ask their IT administrator (the configured path is in its details) — the app never falls back to another ssh. The connection itself runs through this program on the system-OpenSSH engine, which the SSH connection engine setting's `system-openssh` value selects on every platform, including Windows; when the app's built-in SSH library makes the connection instead, this key still governs configuration evaluation (`ssh -G`), host-key lookups (`ssh-keygen`) and the terminal. On a device that a Claude.ai sign-in has moved to an admin-console organization (see `disableDeploymentModeChooser`), the device profile's value is ignored: set the key in the admin console. There, until the organization's configuration has loaded, no client is pinned.
541547 </Accordion>
542548
543549 <Accordion title="sshTransport details">
from line 553
547553
548554 `auto` or unset: the build's default engine.
549555
550 An explicit value overrides the default at each host's next connect or reconnect. An open app takes a device profile's change at its next configuration check. From a configuration server it takes only `builtin`, once two checks in a row return it (about 12 minutes by default); other values wait until it is reopened or the user signs in again. A value that cannot be read, such as a misspelling, counts as `builtin` and waits likewise. Once a server's configuration has loaded it decides, even if it omits the key; until then the device profile's value applies.
556 An explicit value overrides the default at each host's next connect or reconnect. An open app takes a device profile's change at its next configuration check. From a configuration server it takes only `builtin`, once two checks in a row return it (about 12 minutes by default); other values wait until it is reopened or the user signs in again. A value that cannot be read, such as a misspelling, counts as `builtin` and waits likewise. Once a server's configuration has loaded it decides, even if it omits the key; until then the device profile's value applies. On a device that a Claude.ai sign-in has moved to an admin-console organization (see `disableDeploymentModeChooser`), the device profile's value is ignored: set the key in the admin console. There, until the organization's configuration has loaded, the build's default engine is used.
551557 </Accordion>
552558</AccordionGroup>
553559
from line 702
696702 </Accordion>
697703
698704 <Accordion title="coworkEgressAllowedHosts details">
699 Applies only to **tool calls**, in Cowork, Code and Chat. In Cowork it covers the sandbox's web fetch, shell commands and package installs. In Chat it covers web fetch, and the file-analysis sandbox only if you open its network. In Code it's [translated into Claude Code's network sandbox allowlist](https://claude.com/docs/third-party/claude-desktop/code#applied-as-managed-policy), and by default a separately deployed Claude Code managed-settings file takes precedence. It does **not** cover Web Search (which runs at your inference provider), inference, or MCP traffic. A list other than `*` also bounds where Claude may browse in the built-in browser. When unset, Cowork and Chat refuse web fetch.
705 Applies only to **tool calls**, in Cowork, Code and Chat. In Cowork it covers the sandbox's web fetch, shell commands and package installs. In Chat it covers web fetch and commands in the file-analysis sandbox. In Code it's [translated into Claude Code's network sandbox allowlist](https://claude.com/docs/third-party/claude-desktop/code#applied-as-managed-policy), and by default a separately deployed Claude Code managed-settings file takes precedence. It does **not** cover Web Search (which runs at your inference provider), inference, or MCP traffic. A list other than `*` also bounds where Claude may browse in the built-in browser. When unset, Cowork and Chat refuse web fetch.
700706
701707 Entries are exact hostnames (`api.github.com`), wildcards (`*.corp.com` matches subdomains at any depth, not `corp.com` itself), or `*` to allow all. IP addresses match only when listed exactly. In Cowork and Chat, web fetch always blocks `localhost` and private addresses, and shells reach only listed hosts (not your inference provider's, unless listed), or none when unset. Code's shell obeys that limit only on macOS, and with bubblewrap and socat on Linux or WSL2; elsewhere, Windows included, it reaches any host. Code's web fetch isn't limited.
702708
from line 834
828834 </Accordion>
829835
830836 <Accordion title="mcpPersistentAlwaysAllowEnabled details">
831 When enabled (the default), approval prompts for tools without a `toolPolicy` entry offer a persistent grant — **Always allow**, or **Allow for all tasks** for tools that can modify data — the Tool permissions picker in Connector settings lets users pre-approve tools, and those grants persist across sessions with no expiry.
837 When enabled (the default), approval prompts for tools without a `toolPolicy` entry offer a persistent grant: **Always allow**, or **Allow for all tasks** for tools that can modify data. The Tool permissions picker in Connector settings also lets users pre-approve tools. These grants persist across sessions with no expiry.
832838
833 When disabled, the persistent options are hidden from approval prompts and from the Connector settings picker, and previously stored persistent grants stop being honored. Scheduled-task runs also stop offering, recording and replaying the **Allow for all scheduled runs** approval, unless a scheduled-task approval lifetime of one day or more is set. Users can still approve each call. Outside scheduled-task runs they can also choose **Allow for this task**, except for read-only tools on servers they added themselves.
839 When disabled, the persistent options are hidden from approval prompts and from the Connector settings picker, and previously stored persistent grants stop being honored. Scheduled-task runs also stop offering, recording, and replaying the **Allow for all scheduled runs** approval (**Always allow** in Code sessions), unless a scheduled-task approval lifetime of one day or more is set. Users can still approve each call. Outside scheduled-task runs, in chat and Cowork they can also choose **Allow for this task**, except for read-only tools on servers they added themselves.
834840
835841 A per-tool `toolPolicy` entry on `managedMcpServers` always takes precedence over this key: `blocked`, `ask`, and `allow` behave exactly as documented there whether this key is enabled or not.
836842
837 This key governs the chat and Cowork surfaces. Code sessions use a separate permission path this key does not cover — govern Code tool approvals with per-tool `toolPolicy` entries, whose `blocked` and `ask` values are forwarded there.
843 This key governs chat and Cowork, and also Code sessions on the device (including WSL) or over SSH. In those Code sessions, with the key disabled, the prompt for an MCP tool offers **Allow for this session** instead of **Always allow**, except on scheduled-task runs. This approval is not saved to Claude Code's settings files. It lasts until the session is deleted, and sessions forked or started from it do not inherit it. Claude Code's own allow rules and hooks still apply. Allow rules in its settings files, including ones saved earlier with **Always allow**, and hooks in its settings or plugins keep allowing the tool until they are removed. Code sessions receive only the `blocked` and `ask` values of `toolPolicy` entries, so a tool set to `allow` still prompts there, and this key applies to it.
838844 </Accordion>
839845
840846 <Accordion title="mcpScheduledTaskApprovalLifetimeDays details">
from line 848
842848
843849 The limit covers MCP tools only, and a `toolPolicy` entry on `managedMcpServers` still takes precedence. If a configuration server delivers the setting, then after each launch scheduled tasks ask instead of reusing stored approvals (as at `0`) until the app has fetched the configuration again.
844850
845 Cowork: if `mcpPersistentAlwaysAllowEnabled` is off, this setting brings the lasting option back for scheduled tasks only, with the time limit.
851 If `mcpPersistentAlwaysAllowEnabled` is off, a limit of one day or more brings the lasting option back for scheduled tasks only.
846852
847 Code tab: `mcpPersistentAlwaysAllowEnabled` has no effect. The limit covers only lasting approvals given after this setting took effect. An earlier one was usually saved as an allow rule in Claude Code's own settings: for a scheduled task, the project's `.claude/settings.local.json`, under `permissions.allow`. Any allow rule in Claude Code's settings, including one written by hand or by **Always allow** in an ordinary Code session in the same project, keeps allowing the tool without asking, even at `0`, until that entry is removed.
853 Code tab: the limit covers only lasting approvals given after this setting took effect. An earlier one was usually saved as an allow rule in Claude Code's own settings: for a scheduled task, the project's `.claude/settings.local.json`, under `permissions.allow`. Any allow rule in Claude Code's settings keeps allowing the tool without asking, even at `0`, until that entry is removed. That includes a rule written by hand, and one saved by **Always allow** in an ordinary Code session in the same project, which the app offers whenever `mcpPersistentAlwaysAllowEnabled` is on. In an ordinary Code session over SSH, if this setting has a value (even `0`), **Always allow** for a tool of a server set up on the host lasts for the session only and saves no rule there.
848854 </Accordion>
849855
856 <Accordion title="isLocalDevMcpEnabled details">
857 Set to `false` in a device-management profile, this key also locks down the local (`stdio`) entries of `managedMcpServers`: the app starts one only when that profile itself lists an entry of the same name, and with the profile's own command, arguments and environment. Only the profile's value engages this. A `false` from a configuration server or the admin console turns user-added servers off and does not engage it.
858 </Accordion>
859
850860 <Accordion title="allowedPluginMcpServers details">
851861 Unset (default): with managed servers listed, sessions take MCP servers only from plugins and the desktop; with no managed servers listed, nothing is restricted.
852862
from line 1073
10631073
10641074| Setting | Type | Availability | Default | Description |
10651075| - | - | - | - | - |
1066| <span id="inferencemaxtokensperwindow" />Max tokens per window<br />`inferenceMaxTokensPerWindow` | `integer` | MDM + Bootstrap<br />Added in 1.2581.0 | — | Per-user soft cap, counted client-side over the token cap window. Not a server-enforced quota. |
1067| <span id="inferencetokenwindowhours" />Token cap window<br />`inferenceTokenWindowHours` | `integer` | MDM + Bootstrap<br />Added in 1.2581.0 | — | Tumbling window length for the token cap. Max 720 hours (30 days). Range: 1–720. |
1076| <span id="inferencemaxtokensperwindow" />Max tokens per window<br />`inferenceMaxTokensPerWindow` | `integer` | MDM + Bootstrap<br />Added in 1.2581.0 | — | Limits each person’s input and output tokens per window on each device. Cached tokens do not count, so this does not cap your provider’s bill. |
1077| <span id="inferencetokenwindowhours" />Token cap window<br />`inferenceTokenWindowHours` | `integer` | MDM + Bootstrap<br />Added in 1.2581.0 | — | The count resets to zero at the end of each window. A window can last 1 to 720 hours (30 days). Range: 1–720. |
10681078
10691079<AccordionGroup>
10701080 <Accordion title="inferenceMaxTokensPerWindow details">
10711081 Requires `inferenceTokenWindowHours` to also be set — without a window length the cap is inert and no limit is enforced.
1082
1083 The app keeps each person’s count in a local file, and no server enforces it. Someone who uses two devices has a separate count on each.
10721084 </Accordion>
10731085
10741086 <Accordion title="inferenceTokenWindowHours details">
10751087 Required when `inferenceMaxTokensPerWindow` is set — the cap only takes effect once both are configured.
1088
1089 Each person’s window starts the first time they use the app after their previous window ends, so 24 hours does not mean midnight to midnight.
10761090 </Accordion>
10771091</AccordionGroup>
10781092
third-party/claude-desktop/configuration-changelog Changed · +8 / -0 lines
from line 4
44
55Configuration keys by Claude Desktop release. Each section lists keys added in that release, with the MDM key name (for plist/registry deployment) and the equivalent JSON shape (for local-file or bootstrap remote configuration).
66
7<Update label="v2.42304.0" description="2026-10-15">
8 **Changed:**
9
10 * `chatAdvancedFileAnalysisEnabled` and `coworkEgressAllowedHosts`: commands in Chat's file-analysis sandbox can now reach the listed hosts. With no list, or an empty one, they have no network access.
11 * `mcpPersistentAlwaysAllowEnabled`: `false` now also applies to Code sessions on the device, in WSL and over SSH.
12 * `claudeAiImport.automatic3pImport` now also reinstalls, without asking, the optional plugins that a user had switched on from any optional `allowedPluginMarketplaces` entry with the same source in the old and new organization.
13</Update>
14
715<Update label="v2.37337.1" description="2026-10-13">
816 <div className="cfg-keys">
917 | MDM key | Type | Description |
government/deploy-desktop/configure Changed · +1 / -1 lines
from line 199
199199"disableDeploymentModeChooser"="true"
200200```
201201
202The `.reg` file from the Export menu targets `HKEY_CURRENT_USER`, which is correct for single-machine testing. For fleet deployment, deliver the values under `HKEY_LOCAL_MACHINE` as shown here.
202For fleet deployment, deliver the values under `HKEY_LOCAL_MACHINE` as shown here. If the `.reg` file from the **Export** menu has `HKEY_CURRENT_USER` in the line in square brackets, change it to `HKEY_LOCAL_MACHINE` before you deploy the file.
203203
204204After a user signs in, Claude for Government sends the app the address to use for chat traffic, which the app shows as **Gateway base URL**. When the registry values are under `HKEY_LOCAL_MACHINE`, the app accepts that address without asking the user. When they are under `HKEY_CURRENT_USER`, the app asks each user once to allow it, as described under [Keys that require user consent](/docs/third-party/claude-desktop/bootstrap#keys-that-require-user-consent). Have each user confirm that the address is on your Claude for Government host before they click **Allow**.
205205