About AnchorShell
Infrastructure between AI applications and model providers, with Relay as the first AnchorShell product.
What AnchorShell builds
AnchorShell builds self-hostable and managed infrastructure for AI applications. Its first product is AnchorShell Relay, an LLM gateway between applications or agents and model providers. The official website is anchorshell.com.
What Relay does
Relay routes supported OpenAI-compatible requests through configured model paths. It queues and paces work against limits, tracks provider cooldowns, applies configured fallbacks, supports external HTTP guardrails, and records operational usage and request traces. Queued requests in the standalone runtime do not survive a restart.
Self-host Relay on your own machine, or use hosted Relay with account and team access. Read the product overview, deployment options, and Getting Started.
Source and documentation
Get the official Relay source on GitHub.
The canonical Relay documentation includes hosted setup, self-hosting, settings, and API access. Each technical article has a Markdown representation. Agents can begin with llms.txt.
Published changes belong in the changelog, and versioned release information is collected on Relay releases.
Contact
Use the contact page for product questions. Security concerns should follow the responsible-disclosure policy.