Repository navigation
chore: sync with upstream pingdotgg/t3code main - #153
Conversation
yordis
commented
Oct 25, 2026
- Upstream merged cleanly with no conflicts, so every ledger divergence carries over unchanged and no ledger entry is retired.
- Migrations stay byte-identical to upstream, keeping fork and upstream migration ids from ever colliding.
- No vendored subtree changed upstream, so .repos/ needs no resync in this change.
…gg#24643) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…atches the page (pingdotgg#24644) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…place (pingdotgg#24671) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…ngdotgg#24738) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…ingdotgg#24742) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…he list (pingdotgg#24749) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…pingdotgg#24760) With the Working shelf on, a thread that stops working stays there for 2s. A server auto-settle in that window sends it straight to Settled, so threads no longer flash in the inbox and vanish. "Thread completed" waits the same 2s and fires once. The server now settles fast enough to land inside the window: it reads a pull request when a shell merge or close command ends, sweeps a thread when its background work ends, and runs per-thread settle checks in parallel, one per thread at a time. Takes over pingdotgg#22458. Co-authored-by: Omar <76748259+omar9848492@users.noreply.github.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…gdotgg#24650) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
pingdotgg#24710) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…gdotgg#24766) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…mbers drafts (pingdotgg#24750) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…tgg#24770) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…4696) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…dotgg#24781) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…4784) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…gdotgg#24782) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…gdotgg#24769) Mobile push said "Thread completed" when a run ended and its continuation started a moment later, then again when the work really finished. Every completion now waits the relay's existing 5s confirm window and is dropped if the thread is live again. A thread that auto-settles during the hold still alerts its completion, then clears its card on a new window. Failed runs publish at once. Follow-up to pingdotgg#24760. Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
pingdotgg#24808) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…tgg#24748) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…ingdotgg#24809) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…PR (pingdotgg#24830) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…pt eats the command (pingdotgg#24775) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…tgg#24813) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…miss (pingdotgg#24832) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…ngdotgg#24834) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…pstream-20261025-1 Signed-off-by: Yordis Prieto <yordis.prieto@gmail.com>
PR SummaryHigh Risk Overview CI wires a maintainer-only shared Vite+ remote task cache ( Mobile navigation replaces ad-hoc thread/PR selection ( iOS UI adds Review / PR diffs report viewport row indices against the full row payload when files are collapsed; JS highlighting skips collapsed files and fixes stale collapse work. Source loading drops a hard attempt cap, evicts cancelled shared requests correctly, and PR routes defer heavy diff build until transitions settle, with cached parsed diff pages and adapter reuse. Reviewed by Cursor Bugbot for commit ee5c64d. Bugbot is set up for automated code reviews on this repo. Configure here. |
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes using default effort and found 1 potential issue.
❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, have a team admin enable autofix in the Cursor dashboard.
Reviewed by Cursor Bugbot for commit ee5c64d. Configure here.
| const { detail, sheet } = (action as unknown as SelectWorkspaceDetailAction).payload; | ||
| const next = workspaceDetailState(state, detail, sheet, nativelyPoppingRoute); | ||
| const reset = { type: "RESET", payload: { ...next, stale: false } }; | ||
| return original.getStateForAction(state, reset as unknown as Action, options); |
There was a problem hiding this comment.
Workspace detail routes lack keys
High Severity
workspaceDetailState builds new Thread, PullRequest, and sheet routes without key, then workspaceDetailRouter RESETs with stale: false. React Navigation only rehydrates when staleis notfalse, so those routes never receive keys. Tests hide this by calling getRehydratedState`. The first list selection or reviewers sheet can therefore mount without a route identity, breaking native-stack identity, back, and in-place updates.
Additional Locations (1)
Reviewed by Cursor Bugbot for commit ee5c64d. Configure here.
Thread transfer impact✅ Thread transfer remains within every enforced ceiling.
Baseline: Scenario and decoded snapshot size10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.
Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed. |

