Add ChatGPT subscription LLM support by neubig · Pull Request #744 · OpenHands/agent-canvas

neubig · 2026-05-23T15:19:44Z

Summary

Add ChatGPT subscription auth constants, service wrappers, query/mutation hooks, and MSW mocks for planned agent-server subscription endpoints.
Extend LLM settings/profile UI with API-key vs ChatGPT subscription mode, subscription model selection, and connect/status/logout controls.
Shape subscription conversation/profile payloads to omit API key/base URL and gate conversation start/profile switch on subscription connection status.

Verification

npm run typecheck
npx vitest run __tests__/api/llm-subscription-service.test.ts __tests__/api/agent-server-adapter.test.ts __tests__/routes/llm-settings.test.tsx
npm run check-translation-completeness
npm test
npm run lint
npm run build

Notes

Live ChatGPT subscription verification still requires real browser-login credentials/session and the planned agent-server subscription endpoints.

This PR was created by an AI agent (OpenHands) on behalf of the user.

🐳 Docker images for this PR

• GHCR package: https://github.com/OpenHands/agent-canvas/pkgs/container/agent-canvas

Component	Value
Image	`ghcr.io/openhands/agent-canvas`
Architectures	amd64, arm64
Agent Server	`ghcr.io/openhands/agent-server:1.24.0-python`
Automation	`openhands-automation==1.0.0a5`
Commit	`5ca3e5502c2f5f27f37cad5c5cb9d9fb839caf0a`

Pull (multi-arch manifest)

# Multi-arch manifest — Docker automatically pulls the correct architecture
docker pull ghcr.io/openhands/agent-canvas:sha-5ca3e55

Run

docker run -it --rm \
  -p 8000:8000 \
  ghcr.io/openhands/agent-canvas:sha-5ca3e55

All tags pushed for this build

ghcr.io/openhands/agent-canvas:sha-5ca3e55-amd64
ghcr.io/openhands/agent-canvas:add-llm-subscription-support-amd64
ghcr.io/openhands/agent-canvas:pr-744-amd64
ghcr.io/openhands/agent-canvas:sha-5ca3e55-arm64
ghcr.io/openhands/agent-canvas:add-llm-subscription-support-arm64
ghcr.io/openhands/agent-canvas:pr-744-arm64
ghcr.io/openhands/agent-canvas:sha-5ca3e55
ghcr.io/openhands/agent-canvas:add-llm-subscription-support
ghcr.io/openhands/agent-canvas:pr-744

About Multi-Architecture Support

Each tag (e.g., sha-5ca3e55) is a multi-arch manifest supporting both amd64 and arm64
Docker automatically pulls the correct architecture for your platform
Individual architecture tags (e.g., sha-5ca3e55-amd64) are also available if needed

Companion PRs / temporary pins

Backend / agent-server endpoints: Add LLM subscription auth endpoints software-agent-sdk#3367
Typed client methods: Add LLM subscription client methods typescript-client#178
Temporary dependency pin: @openhands/typescript-client points at git+https://github.com/OpenHands/typescript-client.git#add-llm-subscription-client until the client PR is merged and released.
Temporary live backend ref: use OH_AGENT_SERVER_GIT_REF=add-llm-subscription-endpoints for manual live testing until the SDK PR is merged and released.

This PR description update was created by an AI agent (OpenHands) on behalf of the user.

Issue

Fixes #1059

This PR description update was created by an AI agent (OpenHands) on behalf of Graham Neubig.

Co-authored-by: openhands <openhands@all-hands.dev>

vercel · 2026-05-23T15:19:50Z

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
agent-canvas	Ready	Preview, Comment	May 30, 2026 4:31am

all-hands-bot

Review Summary

🟡 Acceptable with critical issues - The subscription auth feature is well-structured overall, but has fundamental issues that must be addressed before merging.

[CRITICAL ISSUES]

[__tests__/api/llm-subscription-service.test.ts] Test Quality: The core test suite violates repository guidelines by mocking the entire LLMMetadataClient HTTP client instead of testing real code paths with MSW. This means tests won't catch actual regressions when the HTTP layer, auth headers, or server contract changes. Rewrite to use the MSW handlers already available in settings-handlers.ts.
[src/components/features/settings/llm-profiles/llm-settings-local-view.tsx:170] Complexity: The handleSave function has >4 levels of nesting (try-catch → rename if → auth type if-else → subscription model if → API key if-else → base_url if → wasActive if), violating the repository's '3 levels max' guideline. Extract helpers: buildLlmConfigForAuthType(), handleProfileRename(), handleProfileReactivation().
[src/api/llm-subscription-service.ts:78] Type Safety: Unsafe casting (client as unknown as LLMMetadataClientWithTransport).client to access internal transport. If @openhands/typescript-client changes its internal structure, this fails at runtime with cryptic errors. Add runtime validation that the transport exists and has the expected methods before using it.

[IMPORTANT ISSUES]

[src/components/features/settings/llm-settings/openai-subscription-auth-card.tsx:61] User Experience: When polling returns connected: false, an error toast is shown with a "PENDING" message. This is misleading—users think something failed when they just need to wait. Use an info/warning toast or inline status text instead.
[src/components/features/settings/llm-settings/openai-subscription-auth-card.tsx:36] State Management: Users can get stuck in the challenge state if they start device login but close the tab or navigate away. No way to clear the challenge except completing login or logging out (which fails if not connected). Add a "Cancel" button to reset challenge state.
[src/routes/llm-settings.tsx:219] Data Flow: Model selection is silently changed when switching auth types. If a user has 'gpt-4o', switches to subscription to explore, then switches back, their original selection is lost forever (replaced with defaultModel). Store previous models in component state and restore them, or at minimum show a toast explaining the change.
[src/api/llm-subscription-service.ts:109] Validation: Missing validation for empty strings in normalizeDeviceChallenge. A buggy server could return empty strings that pass the !deviceCode check. Add .trim().length > 0 validation for deviceCode, userCode, and verificationUri.
[__tests__/api/llm-subscription-service.test.ts] Test Coverage: Missing critical test cases: error handling (401/403, 500, timeouts, malformed responses), validation (incomplete device challenges), end-to-end integration (full device flow), and edge cases (empty strings, null values, concurrent requests).
[src/api/llm-subscription-service.ts:87] Error Handling: Network errors from the HTTP client propagate without context. Wrap errors to provide more specific messages about which subscription operation failed.

[SUGGESTIONS]

[src/api/agent-server-adapter.ts:820] Documentation: Add docstring to assertSubscriptionAuthReady noting that it's NOT called when resuming conversations or sending additional messages, so subscriptions could expire mid-conversation. The agent-server must handle expired tokens gracefully.
[src/api/llm-subscription-service.ts:136] Data Structure: The expiresAt field tries both string and number fallbacks, including expires_in which is typically a relative duration in seconds, not an absolute timestamp. Mixing relative and absolute time values in the same field could cause confusion. Document the expected behavior or convert relative durations to absolute timestamps.
[src/hooks/query/use-llm-subscription-status.ts:14] Query Configuration: staleTime of 30 seconds seems short for auth status that rarely changes. Consider 5 minutes to reduce unnecessary network requests, especially since mutations properly invalidate this query.

[RISK ASSESSMENT]

🟡 MEDIUM RISK - This PR adds new authentication flows and modifies conversation creation logic. The main risks are:

Untested error paths could cause poor UX when auth fails
Complex nested logic increases maintenance burden
Type safety issues could cause runtime failures with SDK updates

Not blocking auto-merge based on risk alone, but the critical issues above should be addressed.

VERDICT:

❌ Needs rework - Address the critical test quality and complexity issues, then fix the important UX and validation gaps.

KEY INSIGHT:

The subscription auth implementation follows OAuth device flow patterns correctly, and backward compatibility is preserved (API key flows still work). However, the test suite doesn't provide confidence that the integration will work in production, and the complex nested save logic creates maintenance risk. Focus first on rewriting tests to use MSW and refactoring handleSave.

Was this automated review useful? React with 👍 or 👎 to this review to help us measure review quality.
Workflow run: https://github.com/OpenHands/agent-canvas/actions/runs/26336381445

Co-authored-by: openhands <openhands@all-hands.dev>

- Merge /api/llm/subscription/openai/models with chatgpt/ provider models from /api/llm/models so the subscription dropdown stays in sync with the full LiteLLM registry (e.g. includes gpt-5.5) - Add gpt-5.3-codex to the hardcoded fallback constant - Replace hardcoded OPENAI_SUBSCRIPTION_MODELS array in the dropdown with a dynamic useOpenAISubscriptionModels hook (falls back to constant when endpoints are unavailable) - Device code block: remove 'Code: ' prefix, fix black-on-dark contrast by using --oh-surface-primary / --oh-text-primary tokens, add CopyToClipboardButton for one-click copy Co-authored-by: openhands <openhands@all-hands.dev>

LiteLLM's chatgpt/ provider registry does not yet include gpt-5.5, but it is available via ChatGPT Plus/Pro subscription. Add it to the hardcoded fallback so it appears in the dropdown regardless of whether the backend's chatgpt/ entries have been updated. Co-authored-by: openhands <openhands@all-hands.dev>

- Drop /api/llm/subscription/openai/models as a model source; it was added in this PR set and is redundant — LiteLLM's chatgpt/ provider is the single authoritative list - Remove the OPENAI_SUBSCRIPTION_MODELS hardcoded fallback array, OpenAISubscriptionModel type, DEFAULT_OPENAI_SUBSCRIPTION_MODEL, and isOpenAISubscriptionModel type guard — all replaced by the live subscriptionModels from the hook, or a simple pass-through in non-hook contexts (adapter, profile builder) - Update agent-server-adapter test to reflect pass-through behavior Co-authored-by: openhands <openhands@all-hands.dev>

rajshah4 · 2026-05-28T18:06:37Z

I tested the ChatGPT subscription flow locally. The frontend/model-selection parts of this PR helped, but the remaining
conversation-start failure appears to be in software-agent-sdk / Agent Server runtime handling, not the UI.

Observed behavior:

/api/llm/subscription/openai/status reports connected.
/api/llm/subscription/openai/models returns models.
Direct calls to the ChatGPT Codex endpoint with the stored OAuth token work.
Direct LiteLLM responses and aresponses calls also work with:
- api_base=https://chatgpt.com/backend-api/codex
- model=openai/gpt-5.3-codex
- streaming enabled
- Codex headers
But starting a normal Agent Canvas conversation fails with:

litellm.AuthenticationError: AuthenticationError: OpenAIException - {"detail":"Could not parse your authentication
token. Please try signing in again."}

The likely root cause is that the SDK loses subscription runtime state after serializing/reloading the LLM config.

A saved conversation LLM contains:

{
"auth_type": "subscription",
"subscription_vendor": "openai",
"model": "openai/gpt-5.3-codex"
}

But after validating/reloading it as an SDK LLM, llm.is_subscription becomes False. That means
_get_litellm_api_key_value() does not fetch the ChatGPT OAuth access token, so the ChatGPT Codex endpoint receives invalid/missing auth and returns the token-parse error.

Potential SDK fix:

@Property
def is_subscription(self) -> bool:
return self._is_subscription or (
self.auth_type == "subscription"
and self.subscription_vendor == "openai"
)

There is a second issue after auth is fixed: LiteLLM’s streamed Responses completion can expose completed_event.response as a plain dict, while the SDK expects an object with .output. I verified the stream output can be recovered from response.output_item.done events and parsed successfully.

So I think this PR may need a companion SDK fix for:

Preserving/rederiving subscription mode after LLM config serialization.
Handling dict-shaped streamed Responses completion objects.
Optionally filtering subscription models to ones actually usable by the account.

Co-authored-by: openhands <openhands@all-hands.dev>

neubig · 2026-05-28T21:46:58Z

Updated this stacked PR in 1e94188: merged current main, resolved package conflicts, kept the temporary @openhands/typescript-client branch pin at b025805, rewrote subscription service tests to use MSW/real client paths, tightened device challenge validation, added login cancel/pending UX, preserved model selection when toggling auth type, and refactored profile-save logic. Local verification: npm run lint, npm run build, and full npm test passed.\n\nThis comment was generated by an AI agent (OpenHands) on behalf of neubig.

openhands-ai · 2026-05-28T21:47:43Z

Uh oh! There was an unexpected error starting the job :(

all-hands-bot · 2026-05-28T21:47:53Z

✅ Review complete.

This review was performed through OpenHands Cloud Automation. You can log in and view the conversation here.

all-hands-bot

Review Summary

✅ Previous critical issues resolved — The three blockers from the prior review have all been addressed: tests now use MSW handlers instead of full-client mocks, handleSave complexity has been properly refactored into helper functions (buildProfileLlmConfig, renameProfileIfNeeded, reactivateProfileIfNeeded), and the unsafe (client as unknown as LLMMetadataClientWithTransport).client cast is gone. The cancel button, model restoration refs, and empty-string validation in readString are all present. This is a significant quality improvement.

The OAuth device flow pattern is correctly implemented, subscription credential omission from conversation payloads is good security practice, and withLlmClient's finally { client.close() } is solid resource management.

[IMPORTANT ISSUES]

Unconditional subscription model fetch (src/routes/llm-settings.tsx:140): useOpenAISubscriptionModels() fires on every mount of LlmSettingsScreen, including for users who never use subscription auth. The hook supports { enabled } — pass enabled: isSubscriptionMode where isSubscriptionMode is derived from the current settings' auth type. The silent failure path means no visible bug, but it generates unnecessary traffic for all API-key users and may produce 404/500 noise in agent-server logs.
Race condition: model saved as empty string (src/routes/llm-settings.tsx:408): In buildPayload, if subscriptionModels is still undefined when the user saves (e.g. they switch to subscription and immediately click save before the model list loads), subscriptionModels?.[0] ?? "" evaluates to "". The same race exists in handleAuthTypeChange at line 235 — switching before models load fires onChange("llm.model", ""). The subscriptionModelValue display fallback at line 181–183 hides this in the UI, but the underlying form value is "". Guard the save path: if authType === LLM_AUTH_TYPE_SUBSCRIPTION && !subscriptionModels?.length, show a validation message or block the save.

[SUGGESTIONS]

assertSubscriptionAuthReady missing lifecycle note (src/api/agent-server-adapter.ts:844): Flagged in the previous review, still unaddressed. The function is called at conversation start and profile switch, but not on subsequent message sends or conversation resume. A one-line JSDoc noting this scope prevents future maintainers from assuming token expiry is covered end-to-end.
@openhands/typescript-client git dep tracking (__tests__/package-library.test.ts:57): The comment says "temporarily allowed while this stacked PR waits for the subscription client branch to merge/release" but there is no linked issue or deadline. Link a tracking issue or PR in the comment so this exemption doesn't outlive its purpose.
staleTime of 30s for subscription status (src/hooks/query/use-llm-subscription-status.ts:14): Raised in the previous review. Since both poll and logout mutations already invalidate this query, the 30-second background refetch is redundant. Bumping to 5 minutes reduces background requests with no UX regression.

VERDICT: 🟡 Close to merge-ready. The race condition in buildPayload/handleAuthTypeChange can silently produce a profile with an empty model field — worth a targeted fix before merge. The unconditional model fetch is a UX polish issue. Everything else is non-blocking.

This review was generated by an AI agent (OpenHands) on behalf of the user through OpenHands Automation. View conversation

Co-authored-by: openhands <openhands@all-hands.dev>

openhands-ai · 2026-05-28T21:54:36Z

@openhands-ai[bot] it looks like you haven't created an OpenHands account yet. Please sign up at OpenHands Cloud and try again.

openhands-ai · 2026-05-28T21:55:22Z

@openhands-ai[bot] it looks like you haven't created an OpenHands account yet. Please sign up at OpenHands Cloud and try again.

Co-authored-by: openhands <openhands@all-hands.dev>

neubig · 2026-05-28T23:11:22Z

CI is green on 13ede8a, all review threads are resolved, and this PR is ready for another review/merge gate pass.\n\nThis comment was generated by an AI agent (OpenHands) on behalf of neubig.

Co-authored-by: openhands <openhands@all-hands.dev>

…support

neubig · 2026-05-29T00:23:30Z

Pushed a follow-up dark-theme contrast fix and merged latest main. CI is green on 6a5799a, and review threads remain resolved.\n\nThis comment was generated by an AI agent (OpenHands) on behalf of neubig.

neubig · 2026-05-30T04:24:20Z

@OpenHands fix merge conflicts

openhands-ai · 2026-05-30T04:24:44Z

I'm on it! neubig can track my progress at all-hands.dev

…support Resolved conflicts: - package.json/package-lock.json: updated @openhands/extensions hash to match main; kept @openhands/typescript-client branch pin (add-llm-subscription-client) needed for subscription feature until typescript-client PR #178 is merged and released - src/mocks/handlers.ts: use workspaces-handlers (renamed in main) with resetMockWorkspaces - src/mocks/mcp-handlers.ts: use main's typed MCPTestResponse version - tests/e2e/snapshots/settings-page.snapshot.spec.ts: use 15000 (no underscore) from main Co-authored-by: openhands <openhands@all-hands.dev>

openhands-ai · 2026-05-30T04:31:37Z

Summary

Request

Fix merge conflicts in PR #744 (add-llm-subscription-support).

What Was Done

✅ Fetched latest main — identified it had advanced to 75e578a0 (v1.0.0-alpha.9), introducing 5 files with merge conflicts.

✅ Resolved all 5 conflicts and pushed — merge commit 5ca3e550 was pushed to origin/add-llm-subscription-support, updating the PR.

Conflicts Resolved

File	Resolution
`package.json`	Updated `@openhands/extensions` to main's hash (`e14f740c`); kept `@openhands/typescript-client` on the `add-llm-subscription-client` branch pin (required for subscription feature until typescript-client PR #178 merges)
`package-lock.json`	Same as `package.json`; preserved `git+https://` URL scheme (not `git+ssh://`) to avoid Vercel build failures
`src/mocks/handlers.ts`	Used main's version with the renamed `workspaces-handlers` (`WORKSPACES_HANDLERS` / `resetMockWorkspaces`)
`src/mocks/mcp-handlers.ts`	Used main's fuller version with the `MCPTestResponse` TypeScript type and explanatory JSDoc
`tests/e2e/snapshots/settings-page.snapshot.spec.ts`	Adopted main's `15000` style (no `_` numeric separator) — purely cosmetic

No Extraneous Changes

Only the 5 conflicted files were modified — all other files in the merge were auto-resolved by git with no manual intervention needed. The subscription feature files (llm-subscription-service.ts, openai-subscription-auth-card.tsx, etc.) were untouched and remain intact in the PR.

openhands-ai · 2026-05-30T04:31:41Z

@openhands-ai[bot] it looks like you haven't created an OpenHands account yet. Please sign up at OpenHands Cloud and try again.

github-actions · 2026-05-30T04:38:37Z

📸 Snapshot Test Report

Warning

Snapshot comparison step crashed (timeout, OOM, or runner error) — diff results below may be incomplete or absent.
Check the CI logs for the full error output (look for the "Run snapshot comparison" step).

✅ 6 snapshots changed — acknowledged via the update-snapshots label. New baselines will be uploaded when this PR merges.

Category	Count
🔴 Changed	6
🆕 New	0
✅ Unchanged	67
Total	73

🔴 Changed snapshots (6)

`changes-tab`

changes-diff-viewer

Expected (main)	Actual (PR)	Diff

`onboarding`

onboarding-step-2-setup-llm

Expected (main)	Actual (PR)	Diff

`settings-page` — 2 snapshots

settings-app-page

Expected (main)	Actual (PR)	Diff

settings-page

Expected (main)	Actual (PR)	Diff

`settings-verification` — 2 snapshots

condenser-settings

Expected (main)	Actual (PR)	Diff

verification-settings-off

Expected (main)	Actual (PR)	Diff

✅ Unchanged snapshots (67)

archived-conversation

conversation-panel-with-archived-badges
conversation-view-archived
conversation-view-sandbox-error

automations

automations-delete-modal
automations-list-active-inactive
automations-no-automations
automations-search-no-results

backends-extended

backend-add-blank-disabled
backend-add-cloud-advanced-open
backend-add-cloud-no-key-disabled
backend-add-cloud-with-key-enabled
backend-add-form-partially-filled
backend-add-invalid-url-disabled
backend-add-local-ready
backend-add-name-only-disabled
backend-add-two-column-layout
backend-add-whitespace-host-disabled
backend-after-switch
backend-cancel-nothing-saved
backend-dropdown-two-backends
backend-edit-prefilled
backend-manage-after-removal
backend-manage-two-listed
backend-remove-cancelled
backend-remove-confirmation
backend-switch-overlay

backends

backend-add-modal
backend-manage-modal
backend-selector-open

changes-tab

changes-deleted-file
changes-empty

collapsible-thinking

reasoning-content-collapsed
reasoning-content-expanded
think-action-collapsed
think-action-expanded

mcp-page

mcp-custom-server-1-editor-open
mcp-custom-server-2-url-filled
mcp-custom-server-3-all-filled
mcp-custom-server-4-installed
mcp-custom-server-editor
mcp-empty-installed
mcp-search-filtered
mcp-slack-install-1-marketplace
mcp-slack-install-2-modal
mcp-slack-install-3-filled
mcp-slack-install-4-installed

onboarding

onboarding-step-0-choose-agent
onboarding-step-1-check-backend
onboarding-step-3-say-hello

projects-workspace-browser

projects-workspace-browser

settings-page

add-backend-modal
analytics-consent-modal
home-screen

settings-secrets

secrets-add-form-filled
secrets-add-form
secrets-after-save
secrets-delete-confirm
secrets-list

settings-verification

verification-settings-on

sidebar

sidebar-collapsed
sidebar-conversation-panel
sidebar-filter-menu

skills-page

skills-empty
skills-loaded
skills-no-match
skills-search-filtered
skills-type-filter

Generated by the Snapshot Tests workflow. This comment was created by an AI agent (OpenHands) on behalf of the repo maintainers.

Add ChatGPT subscription LLM support

06f18ec

Co-authored-by: openhands <openhands@all-hands.dev>

vercel Bot deployed to Preview May 23, 2026 15:20 View deployment

all-hands-bot reviewed May 23, 2026

View reviewed changes

Use typed LLM subscription client

83e343c

Co-authored-by: openhands <openhands@all-hands.dev>

vercel Bot deployed to Preview May 23, 2026 15:43 View deployment

github-actions Bot added a commit that referenced this pull request May 23, 2026

snapshot images for PR #744 run 26336866889

1b115d5

neubig added the update-snapshots Intentional snapshot changes — CI diff check bypassed; new baselines uploaded on merge label May 23, 2026 — with OpenHands AI

github-actions Bot added a commit that referenced this pull request May 23, 2026

snapshot images for PR #744 run 26337057495

7946c63

vercel Bot deployed to Preview May 28, 2026 07:11 View deployment

vercel Bot deployed to Preview May 28, 2026 07:15 View deployment

vercel Bot deployed to Preview May 28, 2026 07:24 View deployment

Merge main and address subscription review feedback

1e94188

Co-authored-by: openhands <openhands@all-hands.dev>

vercel Bot deployed to Preview May 28, 2026 21:44 View deployment

neubig added the review-this label May 28, 2026 — with OpenHands AI

all-hands-bot reviewed May 28, 2026

View reviewed changes

all-hands-bot mentioned this pull request May 28, 2026

chore: remove @openhands/typescript-client git pin once subscription client publishes to npm #917

Open

test: link tracking issue #917 to typescript-client git-pin exemption

5780fb7

Co-authored-by: openhands <openhands@all-hands.dev>

github-actions Bot added a commit that referenced this pull request May 28, 2026

snapshot images for PR #744 run 26604200773

5609729

vercel Bot deployed to Preview May 28, 2026 21:54 View deployment

Address subscription settings review feedback

3da19b0

Co-authored-by: openhands <openhands@all-hands.dev>

github-actions Bot added a commit that referenced this pull request May 28, 2026

snapshot images for PR #744 run 26604489018

4caafbd

vercel Bot deployed to Preview May 28, 2026 22:03 View deployment

github-actions Bot added a commit that referenced this pull request May 28, 2026

snapshot images for PR #744 run 26604889639

d42b208

Add workspace mock handlers for snapshots

65e1712

Co-authored-by: openhands <openhands@all-hands.dev>

vercel Bot deployed to Preview May 28, 2026 22:29 View deployment

github-actions Bot added a commit that referenced this pull request May 28, 2026

snapshot images for PR #744 run 26606011126

f79552a

Mock MCP test endpoint in snapshot tests

e5e5bff

Co-authored-by: openhands <openhands@all-hands.dev>

vercel Bot deployed to Preview May 28, 2026 22:47 View deployment

github-actions Bot added a commit that referenced this pull request May 28, 2026

snapshot images for PR #744 run 26606768499

d33a30c

Stabilize home snapshot waits

13ede8a

Co-authored-by: openhands <openhands@all-hands.dev>

vercel Bot deployed to Preview May 28, 2026 23:00 View deployment

github-actions Bot added a commit that referenced this pull request May 28, 2026

snapshot images for PR #744 run 26607259106

2c3b304

Fix subscription settings dark theme contrast

97a8d32

Co-authored-by: openhands <openhands@all-hands.dev>

vercel Bot deployed to Preview May 28, 2026 23:58 View deployment

github-actions Bot added a commit that referenced this pull request May 29, 2026

snapshot images for PR #744 run 26609432630

2264c62

Merge remote-tracking branch 'origin/main' into add-llm-subscription-…

6a5799a

…support

vercel Bot deployed to Preview May 29, 2026 00:11 View deployment

github-actions Bot added a commit that referenced this pull request May 29, 2026

snapshot images for PR #744 run 26609896969

c985c87

vercel Bot deployed to Preview May 30, 2026 04:31 View deployment

github-actions Bot added a commit that referenced this pull request May 30, 2026

snapshot images for PR #744 run 26674530951

6575ed3

neubig mentioned this pull request Jun 3, 2026

Track PR #744: Add ChatGPT subscription LLM support #1059

Open

Conversation

neubig commented May 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Verification

Notes

Companion PRs / temporary pins

Issue

Uh oh!

vercel Bot commented May 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

all-hands-bot left a comment

Choose a reason for hiding this comment

Review Summary

Uh oh!

rajshah4 commented May 28, 2026

Uh oh!

neubig commented May 28, 2026

Uh oh!

openhands-ai Bot commented May 28, 2026

Uh oh!

all-hands-bot commented May 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

all-hands-bot left a comment

Choose a reason for hiding this comment

Review Summary

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

openhands-ai Bot commented May 28, 2026

Uh oh!

openhands-ai Bot commented May 28, 2026

Uh oh!

neubig commented May 28, 2026

Uh oh!

neubig commented May 29, 2026

Uh oh!

neubig commented May 30, 2026

Uh oh!

openhands-ai Bot commented May 30, 2026

Uh oh!

openhands-ai Bot commented May 30, 2026

Summary

Request

What Was Done

Conflicts Resolved

No Extraneous Changes

Uh oh!

openhands-ai Bot commented May 30, 2026

Uh oh!

github-actions Bot commented May 30, 2026

📸 Snapshot Test Report

changes-tab

onboarding

settings-page — 2 snapshots

settings-verification — 2 snapshots

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

neubig commented May 23, 2026 •

edited

Loading

vercel Bot commented May 23, 2026 •

edited

Loading

all-hands-bot commented May 28, 2026 •

edited

Loading

`changes-tab`

`onboarding`

`settings-page` — 2 snapshots

`settings-verification` — 2 snapshots