fix(web): allow setting reasoning effort for scheduled tasks #30253 · pingdotgg/t3code

Before	After saving and reopening
Task editor without effort control	Task editor retaining High effort

MmacroscopeappBotapproved these changes7d ago

Approvability

Verdict: Approved at 540b7fb

Macroscope's review found this PR approvable — Scheduled tasks gain a reasoning-effort picker that saves options with the task’s model selection. Changes remain explicit and task-specific, with existing defaults and save safeguards unchanged.

^{You can add or adjust custom eligibility rules. Learn more.}

CcoderabbitaiBot7d ago

Explain Change →

No actionable comments were generated in the recent review.

Walkthrough

The scheduled-task editor now stores a structured model selection and provides model-trait controls. It submits the selected model options when saving tasks. Browser tests cover creating and editing tasks, changing reasoning effort, and switching to a model without options.

Changes: Scheduled task model traits

Layer / File(s)	Summary
Structured model-selection draft `apps/web/src/components/settings/ScheduledTasksSettings.tsx`, `apps/web/src/components/settings/scheduledTasksSettings.logic.ts`	Draft state and task conversion now use `ModelSelection` directly instead of a model key and separate base selection.
Model traits editing and saving `apps/web/src/components/settings/ScheduledTasksSettings.tsx`, `apps/web/src/components/settings/ScheduledTasksSettings.browser.test.tsx`	The editor shows `TraitsPicker` for the active provider model and updates the structured selection when options change. It submits that selection and clears it when the environment changes. Browser tests cover create and edit flows, including changing reasoning effort and selecting a model without options.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Change: Feature · Severity of issue fixed: Medium

Sequence Diagram(s)

mermaid

1	sequenceDiagram
2	participant User
3	participant ScheduledTasksSettings
4	participant TraitsPicker
5	participant ScheduledTaskUpsert
6	User->>ScheduledTasksSettings: Select provider model
7	ScheduledTasksSettings->>TraitsPicker: Provide model data and current options
8	User->>TraitsPicker: Change model options
9	TraitsPicker->>ScheduledTasksSettings: Return updated options
10	User->>ScheduledTasksSettings: Save task
11	ScheduledTasksSettings->>ScheduledTaskUpsert: Submit modelSelection

Possibly related issues

pingdotgg/t3code#29049 — The PR adds model-option controls to the web scheduled-task editor, which is also part of this issue’s objective.
pingdotgg/t3code#31344 — The PR lets users set effort for a scheduled task, but does not address the issue’s separate concern about options for different-model delegation.

Suggested reviewers: juliusmarminge

Merge Risk: External image Minimal · up to 258e3

No identified issue currently prevents merging the scheduled-task editor change after normal checks.

Pre-merge checks: ^Passed 4

Check name	Status	Explanation
Title check	Passed	The title clearly describes the main change to scheduled-task reasoning effort.
Description check	Passed	The description covers the problem, change, scope, verification, screenshots, and agent details.
Linked Issues check	Passed	[`#30233`] The web editor adds `TraitsPicker` and saves the full `ModelSelection`. Drafts retain configured defaults and saved options. Browser tests cover effort edits and model changes. The PR descrip…
Out of Scope Changes check	Passed	[`#30233`] The web editor changes, draft-state refactoring, and browser tests support the requested effort control and option persistence. The change summary shows no unrelated changes.

Full details: Linked Issues check

Explanation

[#30233] The web editor adds TraitsPicker and saves the full ModelSelection. Drafts retain configured defaults and saved options. Browser tests cover effort edits and model changes. The PR description states that mobile already exposes the controls and saves the full selection.

Finishing Touches

Generate unit tests (beta)

Create a new PR

Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

External image Recent review info

External image Run configuration

Source mode: Auto
Repository: pingdotgg/t3code
Branch: main
File: .coderabbit.config.ts
Commit: 67fdc53
Layers: Repository TS config
Review profile: CHILL
Plan: Advanced
Run ID: `24956561-5417-486b-afed-62e88070c23e`

External image Commits

Reviewing files that changed from the base of the PR and between 8afb8d75eb18e8016819943ee9111f7eaa8229a5 and 258e3007f04e848d31218f0724e7a63c360ef41a.

External image Files selected for processing: 2

`apps/web/src/components/settings/ScheduledTasksSettings.tsx`
`apps/web/src/components/settings/scheduledTasksSettings.logic.ts`

Included review availability: This review used your included allowance. Your plan provides up to 10 included reviews per hour; 1 remain after this review.

Mmaria7d ago

Note

Written by gpt-6.1-sol on behalf of Maria

Verified in a real dev environment

verified head 8afb8d75eb18e8016819943ee9111f7eaa8229a5 as-is on a disposable linux vm, using the real web client in system chrome through playwright. no code changes were needed.

reproduced on github/main at 12ec17ce4c: the scheduled-task editor has no reasoning-effort control.
installed both checkouts with vp i and ran one background dev server at a time, stopping main before starting the pr. created a real git project through the app, with no fixtures or replay provider.
created a paused task using codex gpt-6.1-sol, selected high effort, saved, and reopened it at high. changed high to low and back, saved and reopened both. reselecting the same model retained effort; switching to gpt-6-luna removed saved options, confirmed through a read-only database query.
restored sol/high and exercised both “run now” and a one-time scheduled run at 17:59 utc. both produced new threads and the real response “scheduled effort verified.” in the client. native codex rollout records independently confirm the model, high effort, and project directory.
blacksmith ran vp test run apps/web/src/components/settings/ScheduledTasksSettings.browser.test.tsx apps/web/src/components/settings/scheduledTasksSettings.logic.test.ts: 2 files, 29 tests passed. all 29 existing pr checks were green.

selected runtime transcript:

text

1	manual run: 2026-11-03T17:52:48.210Z model=gpt-6.1-sol effort=high
2	completed: 2026-11-03T17:52:51.621Z "scheduled effort verified."
3	timed run: 2026-11-03T17:59:03.925Z model=gpt-6.1-sol effort=high
4	completed: 2026-11-03T17:59:07.061Z "scheduled effort verified."
5	cwd=/home/user/effort-verification
6	scheduled task: run_count=2 last_run_status=succeeded last_run_error=null

full selected runtime transcript

main: matching task with no reasoning-effort control

pr: saved task reopened with high effort

effort edits, saving, reopening, and same-model reselection

real timed codex response in the client with high effort

Not verified

native ios/android clients, the electron shell, macos/windows, other providers, and relay/tunnel connections were not exercised. the real runtime check used codex sol/high; default-option initialization is covered by the focused browser tests. uploaded video playback and seeking were checked in chrome; the same recording is embedded above as an animated gif for github.

Nnoahflk7d ago

Added 1 commit3d ago

fix(web): allow setting reasoning effort for scheduled tasks258e300

fix(web): allow setting reasoning effort for scheduled tasks#30253

Problem

Change

Scope and approval

Verification

Checks

Approvability

Sequence Diagram(s)

Verified in a real dev environment

Not verified