Repository navigation
Pick each bot's permission mode in its chat - #17
Merged
Merged
Conversation
A bot's avatar is now just the mascot in its color, with a thin edge (dark on the light theme, light on the dark one) so light colors such as the palette's beige keep their shape. Botloft's own white mascot, in the title bar and on the daemon's messages, keeps the black square of the app icon: without it, it would vanish on the light theme.
The whole window is now drawn at 125% (the webview's zoom), so text, spacing, icons and avatars grow together; at 100% the app looked tiny on a monitor without Windows scaling. A Size menu in the title bar picks 100%, 110%, 125% or 150%, Ctrl+= and Ctrl+- step through them and Ctrl+0 goes back to the default. The choice is remembered and applied before the first paint, so the window never flashes small.
Like the account corner of a chat app: the owner's name (the Windows account's display name), their Claude plan and a menu with Usage, Settings, Language, What's new and Help. Usage shows each window of the plan the bots have used and when it resets; Settings holds the theme, size and language, and the versions. Theme, size and language leave the title bar, which keeps only the language on the setup screens that have no sidebar. The sidebar now shows from the welcome screen on. The daemon reads the owner's display name once (GetUserNameExW) and the email, plan and organization from `claude auth status`, and sends them in `system.status.account`.
Buttons, fields and menu rows get 8 px corners, cards and menus 12 px, the chat's bubbles, cards and dialogs 16 px. Menus and dialogs float on a light themed shadow. The chat is a centered column with the day as a pill, and the open conversation in the sidebar is an inset rounded block instead of a full-width bar. Menus share one popover look and one hook that closes them.
Like Claude Code's mode picker, below the chat: Auto, Manual, Accept
edits and Plan, each described by what the bot does ("Scout decides
what needs your OK"), and Bypass permissions behind a confirmation that
says the bot is then not held to its folder. A bot that asks nothing is
marked in red in the picker and its header.
The daemon stores the mode (migration 0004) and starts the bot with it.
Claude Code reads --permission-mode only at start, so a new mode
restarts the process with --resume once nothing is in progress, and
never cuts a turn short. When an approved plan moves the bot out of plan
mode, the system/init of the next turn says so and the stored mode
follows.
A bot in plan mode asks to go ahead with ExitPlanMode: the chat shows
the whole plan (kept up to 32 KB) with Approve plan and Keep planning,
which sends back what should change.
Spec: 7.4, 10.1, 11.2, 12, 13 (what bypassing means on Windows), 15.1,
15.3 and two rows in section 19 to check with the real Claude Code.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Each bot gets a permission mode, picked below its chat like Claude Code's mode picker. Stacked on #16; once that merges with a merge commit, this diff shrinks to its own commit.
What changes
--permission-modeonly at start, so a new mode restarts the process with--resumeonce nothing is in progress: at once when idle, or when the running turn and its approvals end. The composer says so while the bot is busy.ExitPlanMode. The chat shows the whole plan (its input is kept up to 32 KB) with Approve plan and Keep planning, which sends back what should change. When an approved plan moves the bot out of plan mode, thesystem/initof the next turn says so and the stored mode follows.bots.permission_mode); protocol:bots.setPermissionModeandBot.permissionMode.Spec: 7.4, 10.1, 11.2, 12, 13 (what bypassing means on Windows), 15.1, 15.3 and section 19.
Tested
cargo fmt,clippy -D warnings,cargo test(newpermission_modestests withFakeRuntime: the flag at start, restarting idle and busy bots, leaving plan mode, an old process not undoing the owner's choice) andpnpm check(mode picker, bypass confirmation, busy note, plan approve and keep planning).system/initcarriespermissionMode.Manual test with the real Claude Code
-p, and which mode the bot is in after approving (spec 19, "ExitPlanModeem-p").--permission-mode autodoes (spec 19, "Modos de permissão em-p").