Documentation

GoToMesh is a deterministic, self-refreshing engine for programmatic content — sites where pages are generated from data and a template and wired into a topic-cluster mesh, rather than hand-written one at a time. gtmesh is its command-line tool.

This guide explains what it does, the thinking behind it, and how to run it end to end. It’s written for a semi-technical founder or marketer — you don’t need to be an engineer.

The problem it solves

Programmatic content is easy to build once and painful to keep alive. The usual approach — an LLM and a spreadsheet — falls apart on the second run:

• The page set is non-deterministic: regenerate it and nothing lines up, so there’s nothing stable to diff.
• A refresh mutates the live site and hopes — you find out what changed after it has already changed.
• The internal-link mesh rots: hand-wired links cannibalise, dangle, and go stale as the site grows.
• Editorial judgment lives as one-off hand-edits, lost on the next run.

GoToMesh fixes this by treating a content mesh the way good engineering teams treat infrastructure: declarative desired state, reviewable diffs, and a mesh that derives itself. You change an input, review the diff, and apply it — the same loop every time. (The home page makes the fuller case.)

How it works, in one line

init → operator → pull → plan → apply → promote → write → seal → publish

Everything orbits one committed store — the graph (graph/nodes.jsonl + graph/edges.tsv) — which records every page that could exist and what state each is in. You author your domain into it (entities with gtmesh graph upsert, source-less pages with gtmesh page add); gtmesh plan shows the diff between what the graph should be and what it is; gtmesh apply makes reality match. The engine is deterministic and never calls a model: the one place an LLM runs is a repo-local writer skill, downstream of a diff you’ve already reviewed.

Start here

Read the guide in order, or jump to what you need:

1. Concepts & theory — the why. The proven practices GoToMesh executes (programmatic SEO, topic clusters, topical authority…), the mental models you need to drive it, and the glossary.
2. Create a mesh — the how. The end-to-end walkthrough from an empty directory to a published page — gtmesh init scaffolds the repo, then the repo-local operator skill drives the build loop.
3. Operate a mesh — the day-to-day. The refresh loop, the reconcile/lifecycle reference, and the tuning levers.
4. Features — the lookup reference. A page per capability (the mesh store, the mesh, Ahrefs integration…) and a page per gtmesh command, for when you need one thing without reading the whole guide.

The CLI section is the tool companion: how to install gtmesh and the full command reference.

Want to get hands-on first? Jump straight to the walkthrough. Want the approach first? Start with Concepts & theory.