Repository navigation
Conversation
ApprovabilityVerdict: Approved at Macroscope's review found this PR approvable — Separates task-follower waits from agent steering and releases OpenCode foreground observation when a pending steer meets a newly started subagent. The changes are narrowly scoped to responsiveness, with bounded background requests and existing access controls preserved. You can add or adjust custom eligibility rules. Learn more. |
…non-JSON payloads (pingdotgg#36898)
…ngdotgg#27716) Co-authored-by: PR Batch Tester <agent@local.test> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> Co-authored-by: maria-rcks <maria@kuuro.net> Co-authored-by: Julius Marminge <julius0216@outlook.com>
Co-authored-by: Julius Marminge <julius0216@outlook.com>
Co-authored-by: Julius Marminge <julius0216@outlook.com>
…tgg#26455) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…dotgg#34314) Co-authored-by: Julius Marminge <julius0216@outlook.com>
…pingdotgg#36909) Co-authored-by: Julius Marminge <julius0216@outlook.com>
…dotgg#34196) Co-authored-by: Julius Marminge <julius0216@outlook.com>
…#36915) Co-authored-by: Julius Marminge <julius0216@outlook.com>
…arting (pingdotgg#14424) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: SAPHID <saphid@users.noreply.github.com> Co-authored-by: Julius Marminge <julius0216@outlook.com>
…#36906) Co-authored-by: Grok 4.7 <noreply@x.ai> Co-authored-by: Julius Marminge <julius0216@outlook.com>
Co-authored-by: Julius Marminge <julius0216@outlook.com>
…everything until reload (pingdotgg#34537) Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Co-authored-by: maria-rcks <maria@kuuro.net> Co-authored-by: Julius Marminge <julius0216@outlook.com>
Co-authored-by: Julius Marminge <julius0216@outlook.com>
…pingdotgg#32260) Co-authored-by: Lakshmi Tanmay <lakshmi@voltcrash.com> Co-authored-by: maria-rcks <maria@kuuro.net> Co-authored-by: Julius Marminge <julius0216@outlook.com>
…mmit (pingdotgg#34662) Co-authored-by: Julius Marminge <julius0216@outlook.com>
…g#36887) Co-authored-by: Grok 4.7 <noreply@x.ai> Co-authored-by: Julius Marminge <julius0216@outlook.com>
Co-authored-by: Julius Marminge <julius0216@outlook.com>
pingdotgg#36922) Co-authored-by: Julius Marminge <julius0216@outlook.com>
…ngdotgg#30581) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> Co-authored-by: Julius Marminge <julius0216@outlook.com>
…approve (pingdotgg#29692) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> Co-authored-by: Julius Marminge <julius0216@outlook.com>
…in the foreground (pingdotgg#29782) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…fatal HTTP 500 screen (pingdotgg#34918)
…d bound (pingdotgg#35937) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…ice hub is off (pingdotgg#32542) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…otgg#36199) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…not the parent mode (pingdotgg#27334) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…gdotgg#27871) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…ngdotgg#35837) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…ut a silent first address (pingdotgg#35546)
…ning (pingdotgg#31548) Co-authored-by: Julius Marminge <julius0216@outlook.com>
Co-authored-by: Julius Marminge <julius0216@outlook.com>
…g#29538) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…ir install folder (pingdotgg#35840) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…es at that model's rate (pingdotgg#32219) Co-authored-by: Julius Marminge <julius0216@outlook.com>
pingdotgg#36616) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> Co-authored-by: GPT 6.1 Sol <codex@openai.com> Co-authored-by: Julius Marminge <julius0216@outlook.com>
|
This PR changes the test suite by +1120 test cases (browser +103, unit +555, server +462). New browser test files: Keep tests that would fail for a real reason, and test logic below the UI where it can be. See Testing. |
|
Note Written by merged current main without rebasing and resolved the peer-wait conflicts, preserving main's 45-second limit and the distinction between agent waits and task-follower waits. updated the context budget to the measured 49,732 bytes. verified the test-delta report includes main's tests because this stacked pr still targets |
Two follow-ups on top of #35641. Both were fixed on #35622 after the commit this branch started from; these are smaller versions written for your code. One commit each, so you can merge this PR or cherry-pick either one.
1. A steer to a parent can complete a remote delegated task early (05918e3, +9/−4 + test)
RemoteDelegation's background follower callsPeerForwarding.waitForThreadwith the parent's scope, so it also watches the parent's steering. Steps:completed,timedOut: true.completedwithtimedOut: falseforced on, so the task completes early with the wrong reply.Fix: only the agent-facing
t3_thread_waityields to input (yieldToInput = falseby default,truefromcall).2. OpenCode: a foreground subagent started after an accepted steer keeps the steer waiting (a9c6af6, +16/−12 + test)
steerTurnchecks for active foreground calls only once, when the steer is accepted. If the model is still generating and then starts a foregroundsubagentin the same step,session.backgroundis never sent. The steer sits in the inbox for the child's whole run.Fix: the inline call moves into
releaseForeground, which is also called fromsession.tool.calledwhen a pending steer meets a new foreground call. It's forked because events are handled under the session lock. It also gets theREQUEST_REPLY_TIMEOUTthe other requests in the file use; this was the only request without one.Tests: the existing steering test now runs five variants, including "child starts after steer" and "background fails". There's a new PeerForwarding test for the follower case. Without the fixes, all four new cases fail; with them, the PeerForwarding and OpenCode2 adapter tests pass (182/182), and
check:changedpasses. The branch includes a merge of your current head (55d338a); the only conflict was the/backgroundrequest assertion frommain.The forwarded-wait test fixture now releases its blocked remote wait when the test ends, rather than
Effect.never. Without that, the new test hangs on teardown.🤖 Generated with Claude Code