Summary

• Hosted-key tool calls now wait for the per-workspace token bucket to refill instead of returning 429 immediately. Behavior is on for every hosted-key call; BYOK paths short-circuit before acquisition and are unaffected.
• Wait is capped at 5 minutes (well under the 90-min Trigger.dev container ceiling). If the cap is exceeded, falls back to today's 429 result.
• Per-workspace+provider FIFO lock (Redis SETNX, key hosted-queue:{provider}:{workspaceId}) serializes callers within a workspace so the bucket drains predictably. Different workspaces never block each other. No-op when Redis is unavailable.
• On terminal upstream 429 (after the existing 3-attempt exponential backoff), executeWithRetry re-enters the queue, re-acquires a fresh hosted key, and retries once.
• New telemetry: platform.hosted_key.queue_waited and platform.hosted_key.queue_wait_exceeded.

Type of Change

• New feature

Testing

• 17 hosted-key rate-limiter tests pass (4 reworked, 2 new for wait-then-succeed)
• 23 broader rate-limiter tests pass; 73 tools/index tests pass
• bun run lint clean
• bun run check:api-validation:strict passes

Checklist

• Code follows project style guidelines
• Self-reviewed my changes
• Tests added/updated and passing
• No new warnings introduced
• I confirm that I have read and agree to the terms outlined in the Contributor License Agreement (CLA)