refactor(web): extract useDraftOverride + useStoredCredentialPresence shared hooks (LUM-2222, LUM-2223) by devin-ai-integration[bot] · Pull Request #33246 · vellum-ai/vellum-assistant

devin-ai-integration · 2026-06-03T14:53:40Z

Prompt / plan

Phase 1 of LUM-2072 remaining scope: extract shared hooks from the settings/AI domain to DRY up duplicate patterns across multiple consumer files.

Closes LUM-2222 (useDraftOverride), closes LUM-2223 (useStoredCredentialPresence).

What changed

useDraftOverride<T> — generic hook that manages a local draft overlaying a server-derived value, with auto-clear when the server value converges after save + cache refetch. Replaces 4 duplicate draft-state patterns (2× web-search-card, 1× image-gen-card, 1× language-model-card).

useStoredCredentialPresence — TanStack Query hook that checks whether a stored credential exists on the daemon. Centralizes org-readiness gating, retry logic for transient daemon errors, and Sentry error reporting. Replaces 2 identical inline query patterns (web-search-card, provider-editor-modal).

Audit findings (fixed)

Null sentinel in useDraftOverride — The original implementation used null as both "no draft" sentinel and a potential valid value for nullable T. Fixed by wrapping internal state in { value: T } | undefined so null is a valid draft value and undefined is the clear signal.
Redundant isLoading conditions in useStoredCredentialPresence — query.isLoading && !!assistantId && enabled && isOrgReady had 3 redundant guards. TanStack Query v5's isLoading = isPending && isFetching already returns false when enabled is false. Simplified to query.isLoading.
Redundant manual draft clear in language-model-card — setDraftActiveProfile(null) in the save handler was redundant because optimistic onMutate updates the cache, which triggers the auto-clear useEffect in useDraftOverride. Removed for consistency with web-search-card and image-gen-card (which rely entirely on auto-clear).
Dual reconcileFromDaemonConfig memos in web-search-card — Two separate useMemo calls both calling reconcileFromDaemonConfig(daemonConfig) with identical deps. Consolidated into a single memo returning both serverWebSearchMode and serverWebSearchProvider.

Test plan

bun test src/domains/settings/ai/ — 57/57 passing
bun run lint — clean (no new warnings)
bunx tsc --noEmit — clean
CI: 7/7 checks expected green

Link to Devin session: https://app.devin.ai/sessions/4b3b130bbf274e5487da3b4992883093
Requested by: @ashleeradka

… shared hooks (LUM-2222, LUM-2223) DRY 4 duplicate draft-state patterns (useState + useEffect auto-clear) into useDraftOverride<T> — web-search-card (2×), image-generation-card, language-model-card. Also standardizes language-model-card from its inconsistent draftInitialized flag to the same auto-clear pattern. DRY 2 duplicate credential-presence TanStack Query hooks into useStoredCredentialPresence — web-search-card + provider-editor-modal. Both consumers now share a single hook with org-readiness gating, retry logic, and Sentry error reporting. Closes LUM-2222 Closes LUM-2223 Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

The exhaustive-deps rule correctly flags these — even though useState setters are referentially stable, the linter cannot verify that for custom hook return values. Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

devin-ai-integration · 2026-06-03T14:53:45Z

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

Disable automatic comment, CI, and merge conflict monitoring

linear · 2026-06-03T14:53:46Z

LUM-2222

LUM-2223

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: a68d4bcd07

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

chatgpt-codex-connector · 2026-06-03T14:55:29Z

-    }
-    return draftActiveProfile ?? activeProfile;
-  }, [draftInitialized, draftActiveProfile, activeProfile]);
+  const [effectiveActiveProfile, setDraftActiveProfile] = useDraftOverride(activeProfile);


Preserve null as a valid profile draft

When the saved activeProfile is non-null, this hook treats null as “no draft”, but the dropdown also uses null to mean “clear the default profile” (val === "" ? null : val). As a result, selecting the blank/default option immediately falls back to the saved profile, isProfileDirty stays false, and the user cannot save activeProfile: null; the previous draftInitialized flag kept those two states distinct.

Useful? React with 👍 / 👎.

False positive — the old code had the same behavior for null.

Old code path (when val === "" triggers setDraftActiveProfile(null) + setDraftInitialized(true)):

draftInitialized = true → return draftActiveProfile ?? activeProfile → null ?? activeProfile → activeProfile

New code path (same trigger):

draft = null → effective = draft ?? serverValue → null ?? activeProfile → activeProfile

Both fall back to activeProfile identically. The draftInitialized flag only controlled the initial render path (!draftInitialized && activeProfile !== null), not the null-draft case — once initialized, draftActiveProfile ?? activeProfile already collapses null into activeProfile, exactly like useDraftOverride does.

Also, the dropdown options are all non-empty profile name strings — val === "" can't be triggered from the options list. The val === "" ? null : val guard is defensive.

Resolved — no behavioral regression.

devin-ai-integration

✅ Devin Review: No Issues Found

Devin Review analyzed this PR and found no potential bugs to report.

View in Devin Review to see 3 additional findings.

- useDraftOverride: use internal wrapper state so null is a valid draft value for nullable T; undefined is the clear signal - useStoredCredentialPresence: remove redundant isLoading conditions (TanStack Query already gates on enabled) - language-model-card: remove redundant manual draft clear in save handler (auto-clear via useEffect handles convergence) - web-search-card: consolidate two reconcileFromDaemonConfig useMemos into one Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

ashleeradka · 2026-06-03T16:00:42Z

@codex review

chatgpt-codex-connector · 2026-06-03T16:03:03Z

Codex Review: Didn't find any major issues. Chef's kiss.

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

vex-assistant-bot

✦ APPROVE at HEAD ec9daee3

Both prior REQUEST_CHANGES findings cleanly resolved in 0866de5c:

contacts-page.tsx:213 — as ChannelsAvailableGetResponse → satisfies ChannelsAvailableGetResponse. Same compile-time check, no cast. If DEFAULT_CHANNELS ever drifts from the wire shape, TS now errors at the synthesis site instead of silently passing.
contact-channels-section.tsx — replaced as ChannelInfo["id"] with a proper type-predicate pattern: KNOWN_CHANNEL_IDS: ReadonlySet<string> populated from the literal union, isKnownChannelId(value): value is ChannelInfo["id"], then guard-and-narrow (if (!isKnownChannelId(ch.type)) continue; followed by bare id: ch.type). Unknown channel kinds from the daemon now fall through the early-continue branch instead of being asserted into a lying type. Exactly the shape I suggested.

No new as casts on runtime-boundary shapes anywhere in the diff. contacts-gateway.ts (return data.contact; etc.) uses the unwrapped typed SDK response shape via the { 200: ResponseType } generic map — properly typed, not asserted.

Merge gate check:

✅ Vex APPROVED at HEAD
✅ Codex 👍 on PR description
✅ CI green (Lint, Type Check, Build, Test, OpenAPI Spec Check, FlexFrame Lint, Socket Security)
Nighttime — merging per the May 26 rule.

Vellum Constitution — Trust-seeking: the schema-first migration now actually pays off because the runtime boundary stays earned. Generated types, predicate guards, satisfies — no shortcuts that hide drift.

… shared hooks (LUM-2222, LUM-2223) (#33246) * refactor(web): extract useDraftOverride + useStoredCredentialPresence shared hooks (LUM-2222, LUM-2223) DRY 4 duplicate draft-state patterns (useState + useEffect auto-clear) into useDraftOverride<T> — web-search-card (2×), image-generation-card, language-model-card. Also standardizes language-model-card from its inconsistent draftInitialized flag to the same auto-clear pattern. DRY 2 duplicate credential-presence TanStack Query hooks into useStoredCredentialPresence — web-search-card + provider-editor-modal. Both consumers now share a single hook with org-readiness gating, retry logic, and Sentry error reporting. Closes LUM-2222 Closes LUM-2223 Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * fix(web): add useDraftOverride setters to useCallback dep arrays The exhaustive-deps rule correctly flags these — even though useState setters are referentially stable, the linter cannot verify that for custom hook return values. Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * fix(web): fix useDraftOverride null sentinel, clean up dead code - useDraftOverride: use internal wrapper state so null is a valid draft value for nullable T; undefined is the clear signal - useStoredCredentialPresence: remove redundant isLoading conditions (TanStack Query already gates on enabled) - language-model-card: remove redundant manual draft clear in save handler (auto-clear via useEffect handles convergence) - web-search-card: consolidate two reconcileFromDaemonConfig useMemos into one Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> --------- Co-authored-by: ashlee@vellum.ai <ashlee@vellum.ai> Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

ashleeradka and others added 2 commits June 3, 2026 14:51

devin-ai-integration Bot assigned ashleeradka Jun 3, 2026

devin-ai-integration Bot requested a review from ashleeradka June 3, 2026 14:53

chatgpt-codex-connector Bot reviewed Jun 3, 2026

View reviewed changes

devin-ai-integration Bot commented Jun 3, 2026

View reviewed changes

vex-assistant-bot Bot approved these changes Jun 3, 2026

View reviewed changes

vex-assistant-bot Bot merged commit 6f81711 into main Jun 3, 2026
7 checks passed

vex-assistant-bot Bot deleted the devin/1780497732-lum-2222-2223-shared-hooks branch June 3, 2026 16:08

This was referenced Jun 3, 2026

refactor(web): split provider-editor-modal into focused modules (LUM-2224) #33256

Merged

refactor(web): split profile-editor-modal into focused modules (LUM-2225) #33286

Merged

This was referenced Jun 4, 2026

refactor(web): split call-site-overrides-modal, fix audit findings (LUM-2228) #33352

Merged

refactor(web): dissolve reconcileFromDaemonConfig bridge utility (LUM-2239) #33464

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

refactor(web): extract useDraftOverride + useStoredCredentialPresence shared hooks (LUM-2222, LUM-2223)#33246

refactor(web): extract useDraftOverride + useStoredCredentialPresence shared hooks (LUM-2222, LUM-2223)#33246
vex-assistant-bot[bot] merged 3 commits into
mainfrom
devin/1780497732-lum-2222-2223-shared-hooks

devin-ai-integration Bot commented Jun 3, 2026 •

edited

Loading

Uh oh!

devin-ai-integration Bot commented Jun 3, 2026

Uh oh!

linear Bot commented Jun 3, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

chatgpt-codex-connector Bot Jun 3, 2026

Uh oh!

devin-ai-integration Bot Jun 3, 2026

Uh oh!

devin-ai-integration Bot left a comment

Uh oh!

ashleeradka commented Jun 3, 2026

Uh oh!

chatgpt-codex-connector Bot commented Jun 3, 2026

Uh oh!

vex-assistant-bot Bot left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

devin-ai-integration Bot commented Jun 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Prompt / plan

What changed

Audit findings (fixed)

Test plan

Uh oh!

devin-ai-integration Bot commented Jun 3, 2026

🤖 Devin AI Engineer

Uh oh!

linear Bot commented Jun 3, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector Bot Jun 3, 2026

Choose a reason for hiding this comment

Uh oh!

devin-ai-integration Bot Jun 3, 2026

Choose a reason for hiding this comment

Uh oh!

devin-ai-integration Bot left a comment

Choose a reason for hiding this comment

✅ Devin Review: No Issues Found

Uh oh!

ashleeradka commented Jun 3, 2026

Uh oh!

chatgpt-codex-connector Bot commented Jun 3, 2026

Uh oh!

vex-assistant-bot Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

devin-ai-integration Bot commented Jun 3, 2026 •

edited

Loading