[chrarnoldus]

Summary

getInferenceProvider returned a single provider. When a Kilo-exclusive model had more than one entry in inference_provider_restriction, it returned null, so such a model was:

• missing from the models-by-provider snapshot
• listed without inference_provider_ids in the catalog
• skipped by the request-level provider.only / provider.ignore check

It is now getInferenceProviders, which returns every distinct provider that may serve the model:

• the stealth provider for stealth models
• every restriction entry, deduplicated, when there is a restriction
• the direct upstream for non-gateway providers
• an empty list for unrestricted OpenRouter/Vercel models, where the snapshot decides

Callers updated:

• Snapshot sync: lists the model under each provider.
• Catalog: publishes every provider in inference_provider_ids.
• checkExclusiveModelProviderAllowed: allows the request when any provider survives only/ignore, and names every provider in the 403.

Every current exclusive model has zero or one restriction entry, so today's behavior does not change. This is a prep PR for #7291, where public custom LLMs declare several inference providers.

Verification

• No manual testing; this is a refactor with no behavior change for the current model list.

Visual Changes

N/A

Reviewer Notes

Unit tests cover multi-provider allow/deny in checkExclusiveModelProviderAllowed, plus multi-provider and repeated restriction entries in getInferenceProviders.

[kilo-code-bot]

Code Review Summary

Status: No Issues Found | Recommendation: Merge

Executive Summary

The getInferenceProvider → getInferenceProviders refactor is behavior-preserving for every current exclusive model (all have zero or one restriction entry), the prior deduplication and test-cleanup findings are resolved, and no new correctness, security, or memory-leak issues were found in the rebased diff.

Files Reviewed (8 files)

• apps/web/src/lib/providers/openrouter/sync-providers.ts
• packages/web-shared/src/lib/ai-gateway/llm-proxy-helpers.ts
• packages/web-shared/src/lib/ai-gateway/llm-proxy-helpers.test.ts
• packages/web-shared/src/lib/ai-gateway/models.test.ts
• packages/web-shared/src/lib/ai-gateway/providers/kilo-exclusive-model.ts
• packages/web-shared/src/lib/ai-gateway/providers/kilo-exclusive-model.test.ts
• packages/web-shared/src/lib/ai-gateway/providers/openrouter/index.ts
• packages/web-shared/src/lib/ai-gateway/providers/openrouter/index.test.ts

Previous Review Summaries (2 snapshots, latest commit cba96eb)

Current summary above is authoritative. Previous snapshots are kept for context only.

Previous review (commit cba96eb)

Status: No Issues Found | Recommendation: Merge

Executive Summary

The incremental changes resolve both prior suggestions: getInferenceProviders now derives providers from a Set (deduplicating repeated restriction entries) and the test cleanup only splices when indexOf finds the model; no new issues found on the changed lines.

Files Reviewed (3 files)

• packages/web-shared/src/lib/ai-gateway/providers/kilo-exclusive-model.ts
• packages/web-shared/src/lib/ai-gateway/providers/kilo-exclusive-model.test.ts
• packages/web-shared/src/lib/ai-gateway/llm-proxy-helpers.test.ts

Previous review (commit fae4971)

Status: 2 Issues Found | Recommendation: Address before merge

Executive Summary

The getInferenceProvider → getInferenceProviders refactor is behavior-preserving for all current model data and introduces no production memory leak; the only findings are two non-blocking hardening suggestions (undeduplicated restriction slugs, and an unguarded test cleanup splice).

Overview

Severity	Count
CRITICAL	0
WARNING	0
SUGGESTION	2

Issue Details (click to expand)

SUGGESTION

File	Line	Issue
packages/web-shared/src/lib/ai-gateway/providers/kilo-exclusive-model.ts	215	getInferenceProviders does not dedupe inference_provider_restriction; a repeated slug would repeat in the 403 message and can duplicate entries in sync-providers.
packages/web-shared/src/lib/ai-gateway/llm-proxy-helpers.test.ts	189	splice(indexOf(...), 1) removes the last array element when indexOf returns -1; guard with >= 0 like the sibling tests.

Files Reviewed (8 files)

• apps/web/src/lib/providers/openrouter/sync-providers.ts - no issues
• packages/web-shared/src/lib/ai-gateway/providers/kilo-exclusive-model.ts - 1 issue
• packages/web-shared/src/lib/ai-gateway/providers/kilo-exclusive-model.test.ts - no issues
• packages/web-shared/src/lib/ai-gateway/llm-proxy-helpers.ts - no issues
• packages/web-shared/src/lib/ai-gateway/llm-proxy-helpers.test.ts - 1 issue
• packages/web-shared/src/lib/ai-gateway/models.test.ts - no issues
• packages/web-shared/src/lib/ai-gateway/providers/openrouter/index.ts - no issues
• packages/web-shared/src/lib/ai-gateway/providers/openrouter/index.test.ts - no issues

Fix these issues in Kilo Cloud

---

_{Reviewed by deepseek-v4.1-flash-1017 · Input: 58.9K · Output: 12.1K · Cached: 1.1M}

_{Review guidance: REVIEW.md from base branch main}

[chrarnoldus]

Closing: #8091 already landed the same change on main (getInferenceProviders returning every provider, multi-provider checkExclusiveModelProviderAllowed, snapshot and catalog updates), so this PR has nothing left to add. #7291 now builds on main directly.