fix(chat): route through LiteLLM gateway, default to protolabs/smart

[mabry1985]

Why

Ava chat was unusable — Anthropic returned rate_limit_error with a useless "message":"Error" on every call. The chat path was going direct to api.anthropic.com via @ai-sdk/anthropic. The team direction is gateway-first: no Anthropic keys, no direct calls, everything routed through https://api.proto-labs.ai/v1.

This PR yanks the direct-Anthropic path out of chat.

Changes

• apps/server/src/lib/ai-provider.ts — rewritten to use @ai-sdk/openai-compatible pointed at the gateway. Auth chain:

  1. GATEWAY_API_KEY env
  2. OPENAI_API_KEY env (alias — gateway accepts both)
  3. apiKeys.protolabsGateway from settings credentials

  Legacy Claude aliases (sonnet, opus, haiku, anything starting with claude-) are silently remapped to protolabs/smart so the UI's stored "sonnet" defaults don't blow up mid-migration.

• apps/server/src/routes/chat/index.ts — default model alias goes from 'sonnet' → 'protolabs/smart'. Existing fall-back chain (header > body > avaConfig.model > default) is preserved.

• apps/server/package.json — adds @ai-sdk/openai-compatible.

Validation

( cd apps/server && npm run build )
GATEWAY_API_KEY=... GATEWAY_BASE_URL=https://api.proto-labs.ai/v1 \
  node apps/server/dist/apps/server/src/index.js

# server.log:
[AIProvider] Gateway provider initialized: baseURL=https://api.proto-labs.ai/v1 keySource=env.GATEWAY_API_KEY

# POST /api/chat:
data: {"type":"reasoning-start","id":"reasoning-0"}
data: {"type":"reasoning-delta","delta":"The"}
data: {"type":"reasoning-delta","delta":" user"}
... (full stream from protolabs/smart)

Before this change, that same payload returned rate_limit_error and NoOutputGeneratedError. The previous Anthropic-direct path is no longer reachable from /api/chat.

Kept intentionally (for follow-ups)

• Export name getAnthropicModel is unchanged. Renaming touches ~10 call sites with no behavior change. Rename in a follow-up.
• providerOptions.anthropic block in chat/index.ts (extended thinking, contextManagement) stays as-is. The AI SDK ignores provider-specific options when the active provider doesn't match, so these are inert under openai-compatible. Cleanup later.
• Settings UI "Claude" provider tile + Anthropic credential field can be removed in a separate UI-side PR. They're inert now because no code path consumes the Anthropic API key.

Operational note

The server process needs GATEWAY_API_KEY and (optionally) GATEWAY_BASE_URL in its environment. For host-process systemd (the automaker-host.service from PR #3617), drop them into /etc/systemd/system/automaker-host.service.d/override.conf:

[Service]
Environment=GATEWAY_API_KEY=...
Environment=GATEWAY_BASE_URL=https://api.proto-labs.ai/v1

…or pipe from Infisical at deploy time.

[coderabbitai]

Warning

Rate limit exceeded

@mabry1985 has exceeded the limit for the number of commits that can be reviewed per hour. Please wait 17 minutes and 4 seconds before requesting another review.

You’ve run out of usage credits. Purchase more in the billing tab.

⌛ How to resolve this issue?

After the wait time has elapsed, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

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 the trial, open-source and free plans. In all cases, we re-allow further reviews after a brief timeout.

Please see our FAQ for further information.

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: a9bb0177-f23b-4c00-b318-8d297203dd0f

📥 Commits

Reviewing files that changed from the base of the PR and between ebc038e and 37f5c83.

⛔ Files ignored due to path filters (1)

• package-lock.json is excluded by !**/package-lock.json

📒 Files selected for processing (3)

• apps/server/package.json
• apps/server/src/lib/ai-provider.ts
• apps/server/src/routes/chat/index.ts

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/chat-default-to-gateway-protolabs-smart

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[github-actions]

Code Review — ? finding(s)

Async review running parallel to CodeRabbit. Findings are advisory; not all are merge blockers.

protoLabs Code Review Report

• Generated: 2026-05-22T06:41:07Z
• Git head: 35bfc6a6edd9dadd75d06819f6fae0ce999a77df
• Features mapped: 3
• Findings: 0

No findings recorded.

[github-actions]

Code Review — ? finding(s)

Async review running parallel to CodeRabbit. Findings are advisory; not all are merge blockers.

protoLabs Code Review Report

• Generated: 2026-05-22T06:46:36Z
• Git head: e1162fd2cf5e83f71ed9402e5d366d541e596cd8
• Features mapped: 3
• Findings: 0

No findings recorded.