Reading & Delivery

Modeling content is only half the job; the other half is getting it out to the things that read it — browsers, other services, and increasingly language models and agents. This section covers every way content leaves Byline and how to serve it efficiently.

• Client SDK — @byline/client, the typed, DSL-style API for querying and writing documents from outside the admin UI.
• Routing & API — the current transport surface (TanStack Start server functions) and the boundary where a stable HTTP API becomes worthwhile.
• Transports (planned) — how Byline intends to layer framework-agnostic logic under host-specific bindings so the same operations can be exposed over different transports.
• Markdown Export — one-way Lexical-to-markdown rendering, the .md URL surface, and llms.txt for agent consumers.
• MCP Server (planned) — exposing Byline content to AI agents over the Model Context Protocol.
• Caching — CDN edge caching, invalidation strategies, and the optional in-memory data cache.
• Search & Retrieval — the pluggable SearchProvider seam (Postgres full-text search built in, vector / hybrid as external drivers), shipped collection/zone search, hydration, and post-ranking authorization, exposed through the Client SDK surfaces developers build site search on.