Show HN: Loomcycle – a sidecar runtime for AI agents (Go binary, Apache-2.0) (github.com)

Loomcycle has officially launched its version 1.0, a lightweight sidecar runtime designed for AI agents, packaged as a single Go binary (approximately 50 MB) that operates alongside applications rather than within them. This release marks a significant milestone as it achieves a complete feature set with a hardened, distribution-ready product. Key technical features include multi-replica high availability, agent-to-agent interoperability, inbound webhooks, and a variety of memory backends, allowing for flexible integration with existing applications. Importantly, Loomcycle supports major AI systems through built-in inference modes, thus catering to diverse workflows. This release is particularly significant for the AI/ML community as it offers a self-hostable and easily deployable alternative to current market options, which often require embedding specific libraries or relying on managed services tied to a single vendor. With Loomcycle, developers can maintain the programming language of their original applications while leveraging agentic capabilities via simple HTTP or gRPC calls. Furthermore, the robust security features, such as multi-tenant authorization and the ability to pause, resume, or snapshot agent states, enhance usability in both trusted and untrusted environments. This versatility sets Loomcycle apart, fostering greater innovation in AI-driven applications.