feat(inference): OpenAI-compatible /v1 router with user-managed API key#2523
Merged
senamakel merged 10 commits intoMay 23, 2026
Conversation
Expose the /v1 OpenAI-compatible surface as a stable router for local harnesses. The /v1/* endpoints now accept either the per-launch core bearer or a stable user-managed external API key stored in the credentials service. The models list aggregates the routing-configured providers (chat, reasoning, agentic, coding, memory, embeddings, heartbeat, learning, subconscious) plus a sentinel 'openhuman' entry that follows the active routing config. The AI Settings panel grows a new section to set, rotate, and clear the external API key and exposes the base URL plus auth-header guidance.
Contributor
|
Warning Review limit reached
Your plan currently allows 2 reviews/hour. Refill in 5 minutes and 23 seconds. Your organization has run out of usage credits. Purchase more in the billing tab. ⌛ How to resolve this issue?After more review capacity refills, a review can be triggered using the We recommend that you space out your commits to avoid hitting the rate limit. 🚦 How do rate limits work?CodeRabbit enforces hourly rate limits for each developer per organization. Our paid plans have higher rate limits than trial, open-source, and free plans. In all cases, review capacity refills continuously over time. Please see our FAQ for further information. ℹ️ Review info⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (20)
📝 WalkthroughWalkthroughThis PR adds external OpenAI-compatible endpoint authentication with a complete feature implementation: frontend settings UI for API key management, backend authentication middleware for ChangesOpenAI-compatible external endpoint support
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes Suggested labels
Suggested reviewers
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. Comment |
coderabbitai
Bot
added
the
feature
Move all hard-coded UI text in the new OpenAI-compatible endpoint section through useT()/en.ts. Document the rule in CLAUDE.md and AGENTS.md so every new user-visible string in app/src/** must go through the i18n catalog.
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Nitpick comments (1)
app/src/components/settings/panels/AIPanel.tsx (1)
2090-2163: 🏗️ Heavy liftExtract the OpenAI-compatible endpoint UI/logic into a separate module.
This file is already far beyond the repository’s preferred source-file size, and this new card + handlers make it harder to maintain and test. Please split this feature into a focused component/hook (for example
OpenAICompatEndpointCard.tsx+ a small state hook) and keepAIPanel.tsxorchestration-only.As per coding guidelines:
**/*.{js,ts,tsx,jsx}: Prefer files ≤ ~500 lines per source file; split modules when growing to maintain readability and single responsibility.Also applies to: 2483-2500
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@app/src/components/settings/panels/AIPanel.tsx` around lines 2090 - 2163, Extract the OpenAI-compatible endpoint UI and related handlers into a new presentational component (e.g., OpenAICompatEndpointCard) plus a small hook (e.g., useOpenAICompatState): move the entire section JSX and its local state interactions (openAiCompatStatus, openAiCompatBusy, setOpenAiCompatDialogOpen, setOpenAiCompatBusy) and side-effect handler clearOpenAICompatEndpointKey into the new files, keep AIPanel.tsx only orchestrating by importing <OpenAICompatEndpointCard /> and passing props or the hook values/dispatchers; ensure the new hook exposes the status, busy flag, open dialog setter, and a clearKey function that wraps clearOpenAICompatEndpointKey and updates status/busy exactly as currently implemented so behavior remains identical.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@src/core/auth.rs`:
- Around line 265-272: verify_external_inference_bearer currently loads Config
and performs auth-store work even when the supplied bearer is empty; add an
early-return check at the start of verify_external_inference_bearer (e.g., if
supplied.trim().is_empty() { return false; }) so the function short-circuits and
avoids loading Config or performing I/O for blank/missing headers before any
config/auth-store access.
In `@src/openhuman/inference/http/server.rs`:
- Around line 284-297: Normalize any trailing temperature suffix like `@0.7` from
default-model strings before calling push_model: for config.default_model (the
block that reads config.default_model.as_deref()...) and for cloud provider
default_model (inside the loop over config.cloud_providers where
cp.default_model is used), strip a trailing "@<number>" pattern (e.g. regex
"@\\d+(?:\\.\\d+)?$" or equivalent) after trimming the string, then call
push_model with the cleaned ID; ensure you still preserve provider scoping for
cloud entries (format!("{}:{}", cp.slug, cleaned_model)) and only push when the
cleaned model is non-empty.
---
Nitpick comments:
In `@app/src/components/settings/panels/AIPanel.tsx`:
- Around line 2090-2163: Extract the OpenAI-compatible endpoint UI and related
handlers into a new presentational component (e.g., OpenAICompatEndpointCard)
plus a small hook (e.g., useOpenAICompatState): move the entire section JSX and
its local state interactions (openAiCompatStatus, openAiCompatBusy,
setOpenAiCompatDialogOpen, setOpenAiCompatBusy) and side-effect handler
clearOpenAICompatEndpointKey into the new files, keep AIPanel.tsx only
orchestrating by importing <OpenAICompatEndpointCard /> and passing props or the
hook values/dispatchers; ensure the new hook exposes the status, busy flag, open
dialog setter, and a clearKey function that wraps clearOpenAICompatEndpointKey
and updates status/busy exactly as currently implemented so behavior remains
identical.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro
Run ID: db99fbf8-788a-49a4-b06b-9f37821a81d9
📒 Files selected for processing (8)
app/src/components/settings/panels/AIPanel.tsxapp/src/components/settings/panels/__tests__/AIPanel.test.tsxapp/src/services/api/__tests__/aiSettingsApi.test.tsapp/src/services/api/aiSettingsApi.tssrc/core/auth.rssrc/openhuman/inference/http/mod.rssrc/openhuman/inference/http/server.rssrc/openhuman/inference/http/tests.rs
Member
Author
|
Pushed follow-up commits 348bd8c + 652be2a routing the new OpenAI-compat panel strings through i18n and adding the rule to CLAUDE.md + AGENTS.md. Pre-push hook bypassed with |
- Short-circuit empty bearers in verify_external_inference_bearer before loading config to avoid per-request I/O on unauthenticated /v1/* traffic. - Normalize @temperature suffixes for config.default_model and cloud_providers[*].default_model in /v1/models, matching the routing-string path so model IDs dedupe consistently.
i18n-coverage gate failed because the new settings.ai.openAiCompat.* keys only landed in en.ts and en-4.ts. Mirror them into every locale's chunk-4 file so each locale aggregates the same key set; non-English locales fall back to the English value until translated.
Coverage gate (diff-cover ≥ 80%) was failing at 50% because the new OpenAI-compat key controls in AIPanel weren't exercised. Add four tests: - Rotate/Clear chrome rendered when has_api_key is true. - Localized "Unavailable" fallback when status resolution rejects. - Clear-key button invokes clearOpenAICompatEndpointKey and flips back to Set key. - Set-key dialog persists the value via setOpenAICompatEndpointKey and flips the panel to Rotate key.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
/v1/*as a stable OpenAI-compatible router for local harnesses./v1/*requests./v1/models, plus a sentinelopenhumanentry that follows active routing.Problem
The
/v1/*OpenAI-compatible surface was only callable from inside the desktop shell because it required the per-launchOPENHUMAN_CORE_TOKEN. Local agent harnesses (CLIs, IDEs, third-party tooling) need a stable bearer that survives core restarts to treat OpenHuman as an OpenAI-compatible router. Additionally,/v1/modelsonly listed cloud provider defaults plus the local Ollama chat model, missing models reachable through the active routing config.Solution
src/core/auth.rs—/v1/*requests now fall back to a stable bearer stored under theexternal-openai-compatprovider in the credentials/auth service. The internal/rpcsurface still requires the per-launch core token.src/openhuman/inference/http/server.rs—models_handlernow de-dupes via aHashSet, includes a sentinelopenhumanmodel, the configureddefault_model, and every distinct provider referenced by the workload routing config.strip_temperature_suffixstrips numeric@0.7style suffixes when constructing model IDs.app/src/services/api/aiSettingsApi.ts— addsloadOpenAICompatEndpointStatus,setOpenAICompatEndpointKey,clearOpenAICompatEndpointKeyover the existing credentials RPC surface.app/src/components/settings/panels/AIPanel.tsx— adds an OpenAI-compatible endpoint card with base URL display, key status badge, and aProviderKeyDialog-backed rotate/clear flow.Submission Checklist
strip_temperature_suffix. If diff-cover flags gaps, follow-up commit will add tests.Impact
/v1/*;/rpcsecurity unchanged.Related
AI Authored PR Metadata (required for Codex/Linear PRs)
Linear Issue
Commit & Branch
Validation Run
pnpm --filter openhuman-app format:check(will run in CI)pnpm typecheck(will run in CI)Validation Blocked
command:N/Aerror:N/Aimpact:N/ABehavior Changes
/v1/*accepts a stable user-managed API key;/v1/modelslists routing-aware providers.Parity Contract
/v1/*for the desktop shell;/rpcunchanged./v1or/v1/...paths; numeric@<temp>suffix stripping leaves non-numeric suffixes alone.Duplicate / Superseded PR Handling
Summary by CodeRabbit
New Features
Tests