BlueprintLabs

We moved /Slash from single-provider assumptions to a multi-LLM architecture with OpenRouter support.

This month we shipped a major architecture change in /Slash: multi-LLM routing instead of a single-provider dependency. The practical goal was resilience and model choice per task type. Some repository tasks need strong tool reliability, some need faster inference, and some need wider context behavior. We added OpenRouter as a provider option so model selection can be tuned by workflow and by cost envelope. We also standardized request/response adapters so provider-specific differences do not leak through the product surface. What this unlocked for us: - cleaner provider failover behavior - controlled experimentation per feature - easier benchmarking across model families This is one of those foundational changes users do not always see directly, but it makes the assistant much more durable in real usage.

/Slash: Multi-LLM support and OpenRouter rollout