WireNet documentation.

Start here

If you’re new, read What is a spine? first, then walk through Quickstart. You should have a runnable end-to-end spine — entities, a handful of operations, the event log, and a hosted MCP endpoint your agents connect to — by the end of those two pages.

What is a spine?

A small, stable, typed core — five to fifteen entities, a database, and an event log — that everything else touches through typed operations. Agents, web UIs, CLIs, overnight pipelines: they all pass through the same narrow doorway. That discipline is what gives the system audit, revert, replay, and the ability to evaluate any change against real production history.

How the docs are organised

The sidebar on the left is the map. From top to bottom:

• Get started — what a spine is, a 15-minute quickstart, and the concepts you’ll see everywhere else.
• Architecture — the three-layer model, the four disciplines that make it work, the five properties it earns.
• The spine — entities, schema design, the event log, and how to evolve it without breaking production.
• Operations — queries, commands, contracts, inverses.
• MCP endpoint — how to connect Codex desktop, Claude Code, Cowork, and other surfaces to a hosted spine.
• Workflows — composing operations into recipes and skills.
• Audit & replay — querying the log, replaying production history against new prompts, diffing results.
• Platform — managing your spine on platform.wirenet.dev.
• Reference — CLI, TypeScript and Python SDKs, error catalogue, changelog.

Status

These docs are written explainer-first. Pages marked soon in the sidebar are planned but not yet written; we’re shipping the concept pages before the reference pages, on purpose. If a page is missing the content you need today, the field notes on the marketing site cover most of the ground in essay form.

Need help?

Email hello@wirenet.dev or open the in-app help panel from platform.wirenet.dev. Customer spines have a dedicated channel reachable from the platform dashboard.