Conversation
…orktrees - Collapse the two pet model prefs into one petModel picker (Settings -> Canvas Pet) shared by observer and executor; drop the Providers pet section. - Add an autoObserve toggle persisted in .cate/pet.json (defaults on for older files); clicking the idle pet always forces an observe regardless. - Prepare the isolated worktree up-front in the controller (no-op for non-git repos), so create_worktree drops out of the tool surface. - Replace blocking wait_for_terminal with create_terminal/send_keys that wait by default and take background:true to fan out across terminals. - Add a remark tool for ephemeral pet speech bubbles; inject workspace snapshot into the observe prompt so the observer doesn't burn calls rediscovering state. - Fresh-session executor continuation loop with continuation/wake caps to bound cost and recover stalls. - Pet avatar: idle float/hop animation, canvas-area corner docking, world avatar slides in and tracks zoom.
Adapt the Cate Agent work to main's companion->runtime rename (#387): keep the pet-layer features (agent-dir variant for isolated Cate Agent transcripts, local-runtime retryLocal recovery, the CateAgentAvatar) and rewire them onto the runtime naming and APIs.
…to feedback panel
…train feedback width
…row input bar wider than toolbar
…attention glow + notification dot instead of auto-open
…ain; bubblier input grow
…o standard text-sm/text-xs
…bar animates width+height to fit
…ollow-up to the running task
… running todo, not transient activity)
…t just the orchestrator
…job stop; per-job glow clearing
…; topic field on Todo
…terminals, per-card stop/edit/review (no panel background)
…reconcile orphaned runs on restart; epoch-guard editJob restart race; solid job-card bg; edit placeholder; remove dead currentTodoId
…sting); run prompt in chosen worktree; terminal glow uses worktree color (theme accent when none)
…o bottom when multi-line
…ed in worktree color, name on hover (+ track the new selector file)
…hover-expands to name
…k icon + title, worktree-tinted
…), collapses to icon + expands on hover, controls bottom-anchored (don't move)
…r input controls vertically
… tooltip (no inline expand → no flicker/squish)
… pinned left, only padding/width animate — no flicker)
…vered worktree had no record, so color change silently no-op'd and never persisted)
…-line stays centered), derived from measured height
… (no vertical drift on growth)
… centered slot (no size change)
…k starting a second job on an occupied target
…(label/branch/folder via useWorktrees)
…nt + stop re-acquiring) so terminals don't go blank/white under context exhaustion
…e DOM renderer instead of failing to a blank white pane
…y worktrees, resume The executor now acts in two explicit modes, chosen per task: - CANVAS & TERMINAL MANAGEMENT done DIRECTLY with its own tools (no CLI): list_terminals / list_panels / close_panel / focus_panel — so "close my terminals", "bring the editor forward", etc. happen immediately. - CODE WORK still delegated to coding-agent CLIs in visible terminals. Add an `answer` tool: deliver a user-facing text result and settle a question / read-only / management job straight to `done` (no review/merge). The job card renders the kept, selectable output with a Dismiss action. Create the isolated worktree LAZILY on the first create_terminal (named from the executor's short derived topic) instead of up-front, so a pure question or management job never mints a throwaway branch. Resume jobs cut short by an app restart: reconcileOrphans marks them `interrupted`, and the card offers Continue (continueJob → fresh re-grounded executor session) alongside Discard.
…test terminalDom now statically imports WebglAddon (the live-context cap / DOM fallback work), so the dispose-during-creation test — which dynamically imports the real terminalLifecycle → terminalDom — pulled the addon's browser-only UMD wrapper into Node and crashed with `self is not defined`. Stub it like the five sibling xterm addons already mocked here (and like every renderer test that touches this module graph).
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
2 participants
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.
Re-opens the Canvas pet layer work by @Anton-Horn.
The original PR #379 was accidentally merged to main during a batch merge and then reverted (#386). This branch (
pet-layer) was restored at its original tip 09f4458, so all of @Anton-Horn's commits are intact. Continue the work here.cc @Anton-Horn