feat: UI/UX enhancement for model call failures (#1213) by eureka928 · Pull Request #1219 · eigent-ai/eigent

eureka928 · 2026-02-11T02:43:15Z

Related Issue

Description

Implements UI/UX improvements for model call failures to guide users toward fixing configuration issues, as described in #1213.

Backend:

Enrich SSE error events with error_code (invalid_api_key, model_not_found, insufficient_quota) using the existing normalize_error_to_openai_format() classifier
Add lightweight PATCH /provider/{id}/invalidate endpoint to mark a provider as invalid without sending full provider data

Frontend:

Add model error toast with contextual message and "Model Settings" link (persistent for invalid_api_key, 8s for others)
On invalid_api_key errors, automatically call the invalidation endpoint to mark the provider invalid on the server
Fix is_valid field mapping — the backend returns is_vaild (legacy typo) but the frontend was reading is_valid, resulting in the green dot always being wrong
Fix save payloads to send is_vaild: 2 (the correct field name and enum value)
Update status indicators across sidebar, BYOK detail view, and Default Settings dropdown:
- Green dot → configured AND valid
- Red dot → configured but invalid
- No dot → not configured
Show inline "API Key Invalid or Expired" warning on the API key input when a provider is invalidated
Default Settings dropdown trigger switches from green to warning styling when the default model is invalid

i18n:

Add toast message and setting warning keys across all 11 locales

Changes Made

File	Change
`backend/app/service/chat_service.py`	Enrich SSE error with `error_code`
`server/app/controller/provider/provider_controller.py`	Add `PATCH /provider/{id}/invalidate`
`src/api/http.ts`	Add `proxyFetchPatch` helper, `PATCH` method support
`src/types/chatbox.d.ts`	Add `error_code` to `AgentMessage.data`
`src/components/Toast/modelErrorToast.tsx`	New — toast with settings link
`src/store/chatStore.ts`	Show toast on error, call invalidation endpoint
`src/pages/Setting/Models.tsx`	Fix `is_valid` mapping, update status dots, add warnings
`src/i18n/locales/*/chat.json`	New toast message keys (×11)
`src/i18n/locales/*/setting.json`	New warning label keys (×11)

Testing Evidence (REQUIRED)

I have included human-verified testing evidence in this PR.
This PR includes frontend/UI changes, and I attached screenshot(s) or screen recording(s).
No frontend/UI changes in this PR.

What is the purpose of this pull request?

Bug fix
New Feature
Documentation update
Other

Contribution Guidelines Acknowledgement

I have read and agree to the Eigent Contribution Guideline

eureka928 · 2026-02-11T02:43:54Z

@Wendong-Fan @Pakchoioioi @4pmtong @bytecii would you review my PR?
Thank you for your time

eureka928 · 2026-02-20T15:38:57Z

Hi @Douglasymlai would you review this PR?
Thank you

Douglasymlai · 2026-02-20T17:28:37Z

Hi @eureka928 , thanks for contributing this PR.

Quick feedback on the UX for this issue: since an invalid model key blocks the user’s task, the error should be handled inside the chat flow (not just as a notification). It would be better to show an inline error message in the chat with an action, so when the model returns an “invalid key” error, the user can directly jump to Model Settings to fix it.

Right now it presents the error information, but adding a clear CTA (for example, Open Model Settings) would make the flow much smoother.

eureka928 · 2026-02-21T07:12:15Z

Hi @eureka928 , thanks for contributing this PR.

Quick feedback on the UX for this issue: since an invalid model key blocks the user’s task, the error should be handled inside the chat flow (not just as a notification). It would be better to show an inline error message in the chat with an action, so when the model returns an “invalid key” error, the user can directly jump to Model Settings to fix it.

Right now it presents the error information, but adding a clear CTA (for example, Open Model Settings) would make the flow much smoother.

Does this work for you?

Looking forward to hearing from you

eureka928 · 2026-02-24T12:43:12Z

@Douglasymlai @a7m-1st this is ready as well, please review 😉

eureka928 · 2026-02-26T11:44:43Z

@Douglasymlai @4pmtong I think this is also good to merge?

Douglasymlai

Everything looks good. @Wendong-Fan @a7m-1st can do the final check for the merging

eureka928 · 2026-02-28T09:33:02Z

Everything looks good. @Wendong-Fan @a7m-1st can do the final check for the merging

I'd like to follow up on this

a7m-1st · 2026-03-03T12:05:44Z

let me focus on this in a moment !

eureka928 · 2026-03-03T12:46:01Z

let me focus on this in a moment !

I fixed the conflict, would you review and merge?

Pakchoioioi · 2026-03-04T23:54:10Z

This redirect doesn’t meet the requirements; it should go to the model settings page.

bug.mov

eureka928 · 2026-03-05T00:32:43Z

@Pakchoioioi it's updated, that was old version

Screencast.from.2026-03-05.01-31-50.webm

Pakchoioioi · 2026-03-05T00:39:14Z

Also, it would be helpful to keep the original error message from the model printed out as well.

eureka928 · 2026-03-05T00:45:41Z

Also, it would be helpful to keep the original error message from the model printed out as well.

like this?

eureka928 · 2026-03-05T01:06:08Z

@Pakchoioioi

you want like whole error message?

eureka928 · 2026-03-05T12:17:22Z

@a7m-1st is this good to merge now?

eureka928 · 2026-03-06T13:18:11Z

@4pmtong @a7m-1st would you review and merge this as well?

a7m-1st · 2026-03-06T17:19:09Z

src/i18n/locales/en-us/chat.json

  "no-reply-received-task-continue": "No reply received, task continue",
-  "splitting-tasks": "Splitting Tasks",
-  "start-task": "Start Task",
-  "message-cannot-be-empty": "Message cannot be empty",
-  "remove-file": "Remove file",
-  "drop-files-to-attach": "Drop files to attach",
-  "expand-input": "Expand input (⌘P)",
-  "queued-tasks": "Queued Tasks",
-  "remove-queued-message": "Remove queued message",
-  "no-agents-added": "No agents added",
  "open-in-ide": "Open in IDE",
  "open-in-vscode": "Open in VS Code",
  "open-in-cursor": "Open in Cursor",
  "open-in-file-manager": "Open in File Manager",
  "failed-to-open-folder": "Failed to open folder",
-  "task-completed-card-title": "Task completed",
-  "task-completed-card-subtitle": "Automate your task with a trigger"
+  "model-error-go-to-settings": "Model Settings",


Can you add a single commit (to make easy to review) to revert those translation keys?

4pmtong · 2026-03-10T06:02:57Z

Hi @eureka928 Could you help resolve the conflicts?

Use normalize_error_to_openai_format() to classify ModelProcessingError and generic Exception errors, including error_code in SSE payloads so the frontend can take targeted action (e.g. invalid_api_key, model_not_found, insufficient_quota).

Lightweight endpoint to mark a provider as invalid without requiring the full provider payload, used by the frontend when an API key error is detected during a model call.

Add PATCH method support to proxyFetchRequest and expose proxyFetchPatch. Add optional error_code field to AgentMessage.data for typed error handling in the frontend.

Show contextual toast on model call failures with a link to Model Settings. Persistent (Infinity duration) for invalid_api_key errors, 8s for others.

Capture preferredProviderId during chat start. On SSE error events, show model error toast and call the invalidation endpoint when error_code is invalid_api_key.

Fix is_valid field mapping to read is_vaild (legacy typo) from the API response. Fix save payloads to send is_vaild: 2 (VaildStatus.is_valid). Update sidebar, BYOK detail, and default model dropdown to show red indicators for invalid providers and warning styling on the dropdown trigger. Show inline API key error when provider is invalidated.

Add chat keys for model error toast messages (invalid_api_key, model_not_found, insufficient_quota) and setting keys for API key invalid warning and default model unavailable labels across all 11 locales.

- Add Action.error and ActionErrorData to queue system so workforce background tasks can push errors to the SSE stream - Fix "simple answer" error handlers to emit SSE error events instead of silently swallowing exceptions - Add PUT fallback when PATCH /invalidate endpoint is not deployed - Store preferredProvider reference for invalidation fallback Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Show model errors (invalid API key, model not found, quota exceeded) as an inline warning card in the chat message list instead of only a toast. The card includes a contextual description and an "Open Model Settings" button that navigates to /setting/models.

The "Open Model Settings" button navigated to /setting/models which redirected to the settings tab. The models page is actually under the agents tab at /history?tab=agents.

Display the raw error from the model provider below the i18n description so users have the full context for debugging.

Extract message from SDK .body attribute or parse embedded dict, and add break-all to prevent long error strings from overflowing.

eureka928 · 2026-03-10T06:07:08Z

Hi @eureka928 Could you help resolve the conflicts?

@4pmtong this is done, if you don't mind, would you review and merge today?
It was approved and opened for a few weeks

4pmtong · 2026-03-11T06:22:17Z

Hi @eureka928 I noticed that quite a few configurations seem to be missing in the i18n files — it looks like some entries may have been lost during conflict resolution. Could you take another look at that part and make sure everything is intact? Thanks!

I've attached a few screenshots above for reference, though they don't cover all the missing entries — there may be more affected areas worth checking.

Restores setting.json keys (gpt-5.4-name, help-improve-eigent, etc.) and chat.json keys (queued-tasks, splitting-tasks, etc.) across all 11 locales that were dropped when resolving merge conflicts.

eureka928 · 2026-03-11T07:35:39Z

Good catch @4pmtong — these were lost during rebase conflict resolution. I've restored all missing keys:

setting.json (4 keys across all 11 locales): gpt-5.4-name, help-improve-eigent, help-improve-eigent-description, we-only-use-the-essential-data-needed-to-run-your-tasks
chat.json (11 keys across all 11 locales): drop-files-to-attach, expand-input, message-cannot-be-empty, no-agents-added, queued-tasks, remove-file, remove-queued-message, splitting-tasks, start-task, task-completed-card-subtitle, task-completed-card-title

Fixed in commit 6baf1b1.

eureka928 · 2026-03-17T10:11:19Z

HI @Wendong-Fan @4pmtong @a7m-1st I addressed all comments

eureka928 · 2026-03-18T12:22:29Z

@fengju0213 would you merge this PR? I addressed all and got approval

eureka928 force-pushed the feat/model-error-ux branch 2 times, most recently from 83fb9b1 to b0f8259 Compare February 18, 2026 21:37

Wendong-Fan requested review from 4pmtong, Douglasymlai and a7m-1st February 21, 2026 16:36

Wendong-Fan assigned eureka928 Feb 21, 2026

Wendong-Fan added this to the Sprint 16 milestone Feb 21, 2026

Wendong-Fan added frontend Review Required labels Feb 21, 2026

eureka928 force-pushed the feat/model-error-ux branch from b702579 to 26168a1 Compare February 24, 2026 12:42

eureka928 force-pushed the feat/model-error-ux branch from 26168a1 to 93a98ea Compare February 26, 2026 06:14

Douglasymlai approved these changes Feb 26, 2026

View reviewed changes

eureka928 force-pushed the feat/model-error-ux branch from 93a98ea to 92de1c6 Compare March 3, 2026 12:45

eureka928 force-pushed the feat/model-error-ux branch from 92de1c6 to e81b85e Compare March 5, 2026 00:25

a7m-1st reviewed Mar 6, 2026

View reviewed changes

eureka928 and others added 13 commits March 10, 2026 07:05

feat(server): add PATCH /provider/{id}/invalidate endpoint

1b8d3ff

Lightweight endpoint to mark a provider as invalid without requiring the full provider payload, used by the frontend when an API key error is detected during a model call.

feat: add proxyFetchPatch helper and error_code to AgentMessage type

2069159

Add PATCH method support to proxyFetchRequest and expose proxyFetchPatch. Add optional error_code field to AgentMessage.data for typed error handling in the frontend.

feat: add model error toast with settings link

666b715

Show contextual toast on model call failures with a link to Model Settings. Persistent (Infinity duration) for invalid_api_key errors, 8s for others.

feat: show toast and invalidate provider on model errors

b09283c

Capture preferredProviderId during chat start. On SSE error events, show model error toast and call the invalidation endpoint when error_code is invalid_api_key.

feat(i18n): add model error and provider status locale keys

7949f41

Add chat keys for model error toast messages (invalid_api_key, model_not_found, insufficient_quota) and setting keys for API key invalid warning and default model unavailable labels across all 11 locales.

fix: use PUT fallback for provider invalidation and clean up debug logs

c81ebcc

fix: navigate to correct model settings page on error

ae829ce

The "Open Model Settings" button navigated to /setting/models which redirected to the settings tab. The models page is actually under the agents tab at /history?tab=agents.

fix: show original model error message alongside friendly description

6e23717

Display the raw error from the model provider below the i18n description so users have the full context for debugging.

fix: preserve original provider error message instead of generic text

1957859

Extract message from SDK .body attribute or parse embedded dict, and add break-all to prevent long error strings from overflowing.

eureka928 force-pushed the feat/model-error-ux branch from 6afc7b4 to 1957859 Compare March 10, 2026 06:05

eureka928 requested a review from a7m-1st March 10, 2026 06:07

4pmtong added the Waiting for Update label Mar 11, 2026

fix: restore i18n keys lost during rebase conflict resolution

6baf1b1

Restores setting.json keys (gpt-5.4-name, help-improve-eigent, etc.) and chat.json keys (queued-tasks, splitting-tasks, etc.) across all 11 locales that were dropped when resolving merge conflicts.

Conversation

eureka928 commented Feb 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Related Issue

Description

Changes Made

Testing Evidence (REQUIRED)

What is the purpose of this pull request?

Contribution Guidelines Acknowledgement

Uh oh!

eureka928 commented Feb 11, 2026

Uh oh!

eureka928 commented Feb 20, 2026

Uh oh!

Douglasymlai commented Feb 20, 2026

Uh oh!

eureka928 commented Feb 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

eureka928 commented Feb 24, 2026

Uh oh!

eureka928 commented Feb 26, 2026

Uh oh!

Douglasymlai left a comment

Choose a reason for hiding this comment

Uh oh!

eureka928 commented Feb 28, 2026

Uh oh!

a7m-1st commented Mar 3, 2026

Uh oh!

eureka928 commented Mar 3, 2026

Uh oh!

Pakchoioioi commented Mar 4, 2026

Uh oh!

eureka928 commented Mar 5, 2026

Uh oh!

Pakchoioioi commented Mar 5, 2026

Uh oh!

eureka928 commented Mar 5, 2026

Uh oh!

eureka928 commented Mar 5, 2026

Uh oh!

eureka928 commented Mar 5, 2026

Uh oh!

eureka928 commented Mar 6, 2026

Uh oh!

a7m-1st Mar 6, 2026

Choose a reason for hiding this comment

Uh oh!

4pmtong commented Mar 10, 2026

Uh oh!

eureka928 commented Mar 10, 2026

Uh oh!

4pmtong commented Mar 11, 2026

Uh oh!

eureka928 commented Mar 11, 2026

Uh oh!

eureka928 commented Mar 17, 2026

Uh oh!

eureka928 commented Mar 18, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

eureka928 commented Feb 11, 2026 •

edited

Loading

eureka928 commented Feb 21, 2026 •

edited

Loading