Onboarding detect provider from api key #5955

zanesq · 2025-12-02T22:06:27Z

Summary

Resurrecting #5477 from @spencrmartin and integrating the api key detection logic from @DOsinga in #5147

verified working locally

- Add auto_detect.rs module for parallel provider testing - Add /config/detect-provider API endpoint - Add DetectProviderRequest and DetectProviderResponse structs - Update OpenAPI specifications - Backend compiles successfully Part of PR #4950 + #5147 integration

- Add ApiKeyTester component calling /config/detect-provider - Update ProviderGuard with ApiKeyTester in onboarding - Add provider icons and regenerate API client - Show progress and results during detection Integration of PR #4950 + #5147

- Make API Key Tester the only grey container (bg-background-muted) - Arrange Tetrate and OpenRouter side-by-side in grid layout - Add 'Other Providers' section with link to settings - Use transparent backgrounds for provider cards - Improve visual hierarchy and spacing - Add provider icons to each card - Maintain hover effects and transitions Layout now follows: [Quick Setup with API Key] (grey) [Tetrate] [OpenRouter] (transparent, side-by-side) [Other Providers] (transparent)

- Change /settings/providers to /configure-providers - Fixes 'No routes matched location' error in console - Allows 'Other Providers' button to work properly

- Add DESIGN_SUMMARY.md with layout improvements overview - Add INTEGRATION_COMPLETE.md with full integration details - Add ROUTE_FIX_INSTRUCTIONS.md for troubleshooting - Update API client files from regeneration

- Remove integration documentation files (kept locally for reference) - Remove backup files - Keep only production code changes

- Replace 'error: any' with 'error: unknown' - Add proper ApiError interface for type safety - Improve error handling with proper type guards - Fixes ESLint @typescript-eslint/no-explicit-any warning

- Reorder imports alphabetically in config_management.rs - Move auto_detect import to correct position in mod.rs - Remove extra blank lines - Fixes cargo fmt --check validation

Backend changes: - Detect provider from key format first (sk-ant- = Anthropic, sk- = OpenAI, etc.) - Only test the detected provider instead of all providers - Return structured error responses with suggestions - Add DetectProviderError struct with detailed feedback Frontend changes: - Handle new structured error responses - Show provider-specific error messages - Display helpful suggestions when validation fails - Improve loading state to show 'Testing API key...' Fixes issue where wrong keys would test against all providers. Now users get clear feedback about their specific key type.

- Add userInActiveSetup state to track when user is actively testing keys - Prevent ProviderGuard from redirecting when user is in setup flow - Add onStartTesting callback to ApiKeyTester component - Fix useEffect dependencies to include userInActiveSetup - Add frontend rejection of Ollama fallbacks in Quick Setup - Add specific handling for OpenRouter keys (sk-or- format) - Remove suggestions box and test results header for cleaner UI - Add debug logging for troubleshooting Fixes issue where users entering incorrect API keys would see error but then get redirected to existing provider configurations instead of staying on setup screen to retry.

- Add disable_ollama_fallback parameter to DetectProviderRequest - Update auto_detect function to respect fallback disable flag - Update OpenAPI schema and TypeScript types - Prepare for future backend compilation Note: These backend changes require Rust recompilation to take effect. Frontend workarounds are in place until backend can be rebuilt.

- Replace format-based detection with parallel testing approach - Test all providers simultaneously using tokio::spawn for better accuracy - Simplify API response (404 for no match instead of structured errors) - Remove complex Ollama rejection logic - let parallel testing handle it - Should properly detect Anthropic, OpenAI, Google, Groq, xAI, and Ollama - Maintains existing Quick Setup UI with improved backend detection This integrates the robust parallel testing approach from PR #5147 into our Quick Setup onboarding flow for better API key detection.

- Add detect_cloud_provider_from_api_key() function that excludes Ollama - Add /config/detect-cloud-provider endpoint for Quick Setup use - Update ApiKeyTester to use cloud-only detection endpoint - Remove frontend Ollama rejection since backend won't return it - Ensures Quick Setup only works with cloud API providers - Maintains full detection (including Ollama) for other use cases This prevents unwanted Ollama fallbacks in Quick Setup while preserving the parallel testing approach from PR #5147 for better accuracy.

The new /config/detect-cloud-provider endpoint requires backend recompilation. Until then, use the existing /config/detect-provider endpoint with frontend filtering to reject Ollama results in Quick Setup. - Use detectProvider() instead of detectCloudProvider() - Add frontend Ollama rejection back - Maintains cloud-only behavior for Quick Setup - Works with current compiled backend

- Change from module import to specific function imports - Fix function calls to not use module prefix - Resolves compilation error for detect_cloud_provider_from_api_key This allows the backend to compile with the new cloud-only detection endpoint when the Rust code is recompiled.

The Ollama rejection logic was accidentally removed in previous changes. This adds it back to prevent unwanted Ollama fallbacks in Quick Setup. - Reject any Ollama detection results in Quick Setup flow - Show proper error message for cloud providers only - Return early to prevent success flow execution - Maintains focus on cloud API providers for Quick Setup

The pure parallel approach from PR #5147 was testing Anthropic keys against OpenAI API, causing incorrect rejections. This hybrid approach: 1. Detects key format first (sk-ant- → anthropic) 2. Tests the likely provider first with the key 3. Falls back to parallel testing if format detection fails 4. Includes OpenRouter support (sk-or- format) This should correctly route sk-ant- keys to Anthropic API for proper validation while maintaining the robustness of parallel testing for unknown formats. Benefits: - Anthropic keys test against Anthropic API (not OpenAI) - Faster detection (test likely provider first) - Robust fallback (parallel testing if needed) - Supports more providers (OpenRouter)

Since backend hybrid detection requires recompilation, add frontend format detection to provide better error messages and debugging. - Detect key format on frontend (sk-ant- → Anthropic, etc.) - Provide format-specific error messages when validation fails - Add debugging to show detected format and key analysis - Better user guidance based on detected provider format This helps users understand why their keys are failing validation even when the backend is still using the old parallel-only approach.

- Enhanced frontend format detection for all major providers (Anthropic, OpenRouter, OpenAI, Google, Groq, xAI) - Added smart workarounds for backend parallel testing issues - Prevent unwanted redirects during API key testing by maintaining userInActiveSetup state - Reject Ollama fallback in Quick Setup to ensure cloud-only provider detection - Configure providers directly when backend detection fails but format is recognized - Use correct model names: claude-3-haiku-20240307 for Anthropic, anthropic/claude-3-haiku for OpenRouter - Add input validation to prevent console log injection issues - Provide detailed error messages and suggestions for failed key validation - Integrate parallel testing approach from PR #5147 with frontend filtering Fixes the core issue where users entering incorrect API keys would be redirected away from setup screen, now they stay on setup with clear error feedback.

- Replace problematic key generation that could result in null or duplicate keys - Use fallback key pattern: message.id || `message-${index}-${role}` - Ensures unique keys across all message components to prevent React warnings - Fixes: 'Encountered two children with the same key, null' warning This resolves the React warning that appeared when switching between providers or when messages lacked proper IDs.

- Update Anthropic model from claude-3-haiku-20240307 to claude-3-opus-20240229 - Update OpenRouter model from anthropic/claude-3-haiku to anthropic/claude-3-opus - Update success messages to reflect Claude Opus selection - Provides users with the most capable Claude model by default This gives users access to Claude 4's enhanced capabilities for better code generation, analysis, and problem-solving performance.

- Change Anthropic model to claude-sonnet-4-0 (correct model identifier) - Update OpenRouter model to anthropic/claude-sonnet-4-0 - Update success messages to reflect Claude Sonnet selection - Ensures compatibility with the latest Claude Sonnet 4 model This uses the proper model identifier for Claude Sonnet 4 as specified.

- Add OpenAI provider handling to smart workarounds for both Ollama fallback and 404 cases - Set default OpenAI model to gpt-4.1 for optimal performance - Configure OPENAI_API_KEY, GOOSE_PROVIDER=openai, GOOSE_MODEL=gpt-4.1 - Add success message: 'Configured OpenAI with GPT-4.1 model' - Ensures OpenAI keys get proper configuration even when backend detection fails Now all major providers (OpenAI, Anthropic, OpenRouter) have smart workarounds with high-quality default models: GPT-4.1, Claude Sonnet 4, and Claude Sonnet 4 via OpenRouter.

@goose

…0-5147 * 'main' of github.com:block/goose: (243 commits) chore: upgrade npm packages (#5951) feat: ActionRequired (#5897) feat(acp): support loading sessions in acp (#5942) docs: add videos to multi-model page (#5938) docs: promote planning guide (#5934) fix: use a lock to ensure only need to run tunnel just in case multiple go… (#5885) Feat: Added custom headers and toggle keyring CLI options (#5017) Feat/automatic update installation (#5345) fix: Added "Merged consecutive assistant messages" to the acceptable issues for moim injection check (#5933) fix: anthropic provider model fetching (#5932) [MCP-UI] add CSP for images to proxy HTML (#5931) fix: correct typo in blog post (AIMDOEL -> AIMODEL) (#5902) feat: @goose in terminal (native terminal support) (#5887) docs: adding AI-friendly features (#5918) Blog/advent of ai announcement (#5917) Extension selector behind ALPHA flag (#5892) blog: typo fixes (#5896) blog: fixing img url (#5895) blog: MCPs for Developers (#5884) docs: Extension Manager MCP (#5883) ... # Conflicts: # crates/goose-server/src/routes/config_management.rs # crates/goose/src/providers/mod.rs # ui/desktop/openapi.json # ui/desktop/src/api/sdk.gen.ts # ui/desktop/src/api/types.gen.ts # ui/desktop/src/components/ProgressiveMessageList.tsx

zanesq · 2025-12-02T22:10:42Z

ui/desktop/src/components/ProgressiveMessageList.tsx

          return (
            <div
-              key={message.id && `${message.id}-${message.content.length}`}
+              key={message.id || `message-${index}-${message.role || 'unknown'}`}


unrelated fix from @spencrmartin "Resolve React key uniqueness warning in ProgressiveMessageList" looks legit to me

shouldn't that just be message.id?

id can be undefined based on the type.. I think its leftover from when we used to modify messages front end. I'll leave it for now until we can come back to the type and check that its now enforced everywhere. Updated to use message timestamp also.

It needs to be optional since messages only get their id once they get written to the database. if we want to keep this in, why the double || and the appending of the message role?

Got it, I simplified it to key={message.id ?? `msg-${index}-${message.created}`} in the latest commit 👍

michaelneale · 2025-12-02T23:20:04Z

@zanesq lovely - do you think we can as part of this enhance this so it could even detect from (interactive?) env that it has a named key (when I tried that, when it does work, is just delightful as zero config) - if you like I can look at ways of layering it on with this (as it would require some hand testing to make sure it works packaged vs justfile style) - ie use the -l style shell, sniff out 3 of the common key styles and suggest it automatically? (I can add it to this one if you like - or if you would rather a follow on LMK)

zanesq · 2025-12-02T23:22:55Z

@michaelneale sure please go ahead, this branch is fine.. it sounds like a great enhancement and you would be better suited to test it etc. Can always follow up after this merges if you think it will take a while.

* main: fix params not being substituted in activities (#5992) blog: MCP Sampling (#5987) Update Anthropic and Google Gemini models to latest API versions (#5980) docs: chat recall tutorial (#5975) fix: `final assistant content cannot end with trailing whitespace` error from Anthropic (#5967) 5527 multiple file popups (#5905) Groq configure fix (#5833) added sidebar contextual information for firefox (#5433) docs: council of mine MCP (#5979) docs: nano banana extension (#5977) fix: remove prompt change, read model from config (#5976) Enable recipe deeplink parameters for pre-population (#5757)

michaelneale · 2025-12-05T06:36:21Z

@zanesq a good start but 2 missing things we need, and I think this should be a priority PR:

1 - it isn't configuring goose properly - extensions completely empty, even built in ones, so something about how it bootstraps is a very different path:

2 - when it did set it up, say with anthropic, it just picked what i thought was a random model (ie haiku) - not sure of logic of that but should have a recommended one for each maybe? (not sure how that works).

Otherwise - I really like this, could be a great improvement IMO, just those 2

crates/goose/src/providers/auto_detect.rs

DOsinga

Can we revert this to my original implementation? #5147

crates/goose-server/src/routes/config_management.rs

crates/goose/src/providers/auto_detect.rs

DOsinga

some more things I spotted

ui/desktop/src/components/ApiKeyTester.tsx

ui/desktop/src/components/ProviderGuard.tsx

ui/desktop/src/components/ApiKeyTester.tsx

zanesq · 2025-12-05T20:11:22Z

@michaelneale good catch I was able to replicate the missing extensions issue intermittently using a custom GOOSE_PATH_ROOT both with the api key detection and with a custom provider, and the extensions show on refresh sometimes and after re-launching. I think its a pre-existing issue since we didn't touch any related config code in this branch and we aren't doing anything special other than what we did before by loading the provider and model and redirecting to home. I will look into that in another PR.

Regarding the preferred model choice, agreed it would be nice addition. I will look into that in a separate PR. Let me know if you have some ideas of how we could do that. Since it would need to be a hard coded list that might not match user preferences I think the best option would be to show a model chooser dialog when onboarding after selecting a provider?

Co-authored-by: spencrmartin <[email protected]> Co-authored-by: Michael Neale <[email protected]>

DOsinga · 2025-12-05T21:11:16Z

Regarding the preferred model choice, agreed it would be nice addition. I will look into that in a separate PR. Let me know if you have some ideas of how we could do that. Since it would need to be a hard coded list that might not match user preferences I think the best option would be to show a model chooser dialog when onboarding after selecting a provider?

the detection code does return a list of models already, so it shouldn't be too hard to show that in a dropdown after detection maybe?

zanesq · 2025-12-05T21:28:40Z

Yeah was thinking not difficult just where it would fit in the flow, should probably do it for configure providers also not just the auto key detection so basically after any provider is configured for the first time.. will see what I can come up with real quick

…nses-streaming * 'main' of github.com:block/goose: Onboarding detect provider from api key (#5955) Fix PATH on Windows for extensions (#6000) recipe(datahub): Add a recipe for searching & understanding data in DataHub (#5859) fix params not being substituted in activities (#5992)

zanesq · 2025-12-05T22:46:43Z

@DOsinga I think it works well! #6005

* origin/main: feat(mcp): elicitation support (#5965) Onboarding detect provider from api key (#5955) Fix PATH on Windows for extensions (#6000) recipe(datahub): Add a recipe for searching & understanding data in DataHub (#5859) fix params not being substituted in activities (#5992) blog: MCP Sampling (#5987)

* 'main' of github.com:block/goose: (21 commits) Hide recipe icon in empty chat (#6022) docs: provider and model config (#6008) Show modal selector after configuring a provider (#6005) docs: additional mcp sampling resources (#6020) Flutter PR Code Review (#6011) feat(mcp): elicitation support (#5965) Onboarding detect provider from api key (#5955) Fix PATH on Windows for extensions (#6000) recipe(datahub): Add a recipe for searching & understanding data in DataHub (#5859) fix params not being substituted in activities (#5992) blog: MCP Sampling (#5987) Update Anthropic and Google Gemini models to latest API versions (#5980) docs: chat recall tutorial (#5975) fix: `final assistant content cannot end with trailing whitespace` error from Anthropic (#5967) 5527 multiple file popups (#5905) Groq configure fix (#5833) added sidebar contextual information for firefox (#5433) docs: council of mine MCP (#5979) docs: nano banana extension (#5977) fix: remove prompt change, read model from config (#5976) ... # Conflicts: # ui/desktop/src/api/sdk.gen.ts # ui/desktop/src/components/bottom_menu/DirSwitcher.tsx

* 'main' of github.com:block/goose: gov: new LF Projects LLC section (#6027) Cleanup: Remove Recipe Key Flow (#6015) chore(deps): bump mdast-util-to-hast from 13.2.0 to 13.2.1 in /documentation (#5963) remove problematic corrupted woff font (#6006) Added search bar / filtering for recipes (#6019) Hide recipe icon in empty chat (#6022) docs: provider and model config (#6008) Show modal selector after configuring a provider (#6005) docs: additional mcp sampling resources (#6020) Flutter PR Code Review (#6011) feat(mcp): elicitation support (#5965) Onboarding detect provider from api key (#5955) Fix PATH on Windows for extensions (#6000) recipe(datahub): Add a recipe for searching & understanding data in DataHub (#5859) fix params not being substituted in activities (#5992)

* 'main' of github.com:block/goose: (159 commits) Cleanup: Remove Recipe Key Flow (#6015) chore(deps): bump mdast-util-to-hast from 13.2.0 to 13.2.1 in /documentation (#5963) remove problematic corrupted woff font (#6006) Added search bar / filtering for recipes (#6019) Hide recipe icon in empty chat (#6022) docs: provider and model config (#6008) Show modal selector after configuring a provider (#6005) docs: additional mcp sampling resources (#6020) Flutter PR Code Review (#6011) feat(mcp): elicitation support (#5965) Onboarding detect provider from api key (#5955) Fix PATH on Windows for extensions (#6000) recipe(datahub): Add a recipe for searching & understanding data in DataHub (#5859) fix params not being substituted in activities (#5992) blog: MCP Sampling (#5987) Update Anthropic and Google Gemini models to latest API versions (#5980) docs: chat recall tutorial (#5975) fix: `final assistant content cannot end with trailing whitespace` error from Anthropic (#5967) 5527 multiple file popups (#5905) Groq configure fix (#5833) ...

spencrmartin and others added 25 commits October 30, 2025 09:55

fix: Correct route to configure-providers page

0d33d81

- Change /settings/providers to /configure-providers - Fixes 'No routes matched location' error in console - Allows 'Other Providers' button to work properly

docs: Add integration documentation and update API client

4d3d73b

- Add DESIGN_SUMMARY.md with layout improvements overview - Add INTEGRATION_COMPLETE.md with full integration details - Add ROUTE_FIX_INSTRUCTIONS.md for troubleshooting - Update API client files from regeneration

cleanup: Remove documentation and backup files

9e1cca0

- Remove integration documentation files (kept locally for reference) - Remove backup files - Keep only production code changes

fix: Remove TypeScript any type in ApiKeyTester

fe869a9

- Replace 'error: any' with 'error: unknown' - Add proper ApiError interface for type safety - Improve error handling with proper type guards - Fixes ESLint @typescript-eslint/no-explicit-any warning

fix: Apply Rust formatting (cargo fmt)

c5cfcc9

- Reorder imports alphabetically in config_management.rs - Move auto_detect import to correct position in mod.rs - Remove extra blank lines - Fixes cargo fmt --check validation

integrate querying provider

9a9549f

zanesq changed the title ~~Zane/onboarding detect provider~~ Onboarding detect provider Dec 2, 2025

zanesq commented Dec 2, 2025

View reviewed changes

zanesq changed the title ~~Onboarding detect provider~~ Onboarding detect provider from api key Dec 2, 2025

zanesq added 2 commits December 2, 2025 14:29

clippy

8dc17e3

cleanup

4ed4e3d

zanesq requested a review from michaelneale December 2, 2025 23:16

zanesq marked this pull request as ready for review December 2, 2025 23:16

michaelneale self-assigned this Dec 2, 2025

michaelneale reviewed Dec 5, 2025

View reviewed changes

crates/goose/src/providers/auto_detect.rs Outdated Show resolved Hide resolved

DOsinga approved these changes Dec 5, 2025

View reviewed changes

crates/goose-server/src/routes/config_management.rs Outdated Show resolved Hide resolved

crates/goose/src/providers/auto_detect.rs Outdated Show resolved Hide resolved

crates/goose/src/providers/auto_detect.rs Outdated Show resolved Hide resolved

DOsinga reviewed Dec 5, 2025

View reviewed changes

zanesq added 5 commits December 5, 2025 09:30

change to only check provider via api

adc4aea

change to only check provider via api and various feedback

7f2166e

fix onboarding for provider settings page

ec8c8e6

add delay before routing to home after success

4d6d6a8

Update text

ab6cda4

zanesq merged commit 1db4070 into main Dec 5, 2025
17 checks passed

zanesq deleted the zane/onboarding-detect-provider branch December 5, 2025 21:06

zanesq added a commit that referenced this pull request Dec 5, 2025

Onboarding detect provider from api key (#5955)

4abb364

Co-authored-by: spencrmartin <[email protected]> Co-authored-by: Michael Neale <[email protected]>

github-actions bot mentioned this pull request Dec 5, 2025

chore(release): release version 1.16.0 (minor) #5985

Merged

zanesq mentioned this pull request Dec 5, 2025

Show model selector after configuring a provider #6005

Merged

github-actions bot mentioned this pull request Dec 16, 2025

chore(release): release version 1.17.0 (minor) #6131

Closed

Onboarding detect provider from api key #5955

Onboarding detect provider from api key #5955

Uh oh!

Conversation

zanesq commented Dec 2, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Uh oh!

zanesq Dec 2, 2025

Choose a reason for hiding this comment

Uh oh!

DOsinga Dec 5, 2025

Choose a reason for hiding this comment

Uh oh!

zanesq Dec 5, 2025

Choose a reason for hiding this comment

Uh oh!

DOsinga Dec 5, 2025

Choose a reason for hiding this comment

Uh oh!

zanesq Dec 5, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

DOsinga Dec 5, 2025

Choose a reason for hiding this comment

Uh oh!

michaelneale commented Dec 2, 2025

Uh oh!

zanesq commented Dec 2, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

michaelneale commented Dec 5, 2025

Uh oh!

Uh oh!

DOsinga left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

DOsinga left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

zanesq commented Dec 5, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

DOsinga commented Dec 5, 2025

Uh oh!

zanesq commented Dec 5, 2025

Uh oh!

zanesq commented Dec 5, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

zanesq commented Dec 2, 2025 •

edited

Loading

zanesq Dec 5, 2025 •

edited

Loading

zanesq commented Dec 2, 2025 •

edited

Loading

zanesq commented Dec 5, 2025 •

edited

Loading