[diegosouzapw]

Closes #3881

Problem

Adding a valid Nous Research API key reports invalid, even though the exact curl from the portal's copy-shell button works.

Root cause (src/lib/providers/validation.ts)

validateNousResearchProvider probes POST /v1/chat/completions with model: "nousresearch/hermes-4-70b" — a slug Nous does not serve (the registry's real model is Hermes-4-70B). Nous returns 400 model-not-found, and the validator only treated 200/429 as success (with explicit 401/403, 402, >=500 branches), so a 400 fell through to "Connection failed while testing Nous Research" → UI shows invalid. The base URL (https://inference-api.nousresearch.com/v1) and Bearer auth were already correct.

Fix

1. Probe with the real Hermes-4-70B slug (aligns with providerRegistry.ts).
2. Treat any non-auth 4xx (400/404/422) as a valid key — the 4xx proves auth was accepted, only the request shape was wrong. Mirrors the longcat/nvidia validators, so a future model rename can't re-break key validation.

Validation (TDD — Hard Rule #18)

tests/unit/provider-validation-specialty.test.ts:

• Updated the existing "accepts credentials" test to assert the corrected probe model (failed before the source fix).
• New test: a valid Bearer key whose probe returns 400 model-not-found → valid: true (failed before, passes after).

node --import tsx/esm --test tests/unit/provider-validation-specialty.test.ts → 111 pass / 0 fail (was 2 failing pre-fix). typecheck:core clean, lint 0 errors, file-size re-baselined (4394→4407, validator branch not extractable).

[gemini-code-assist]

Warning

You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again!

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.