Scribere Ops

Overview

This skill covers day‑to‑day operations for Scribere and the Semantic Scroll instance, with a focus on safe updates, predictable builds, and instance ownership of templates, assets, and content. It avoids destructive Git actions and keeps changes aligned with the repository specs.

Workflow

Confirm which repo is in scope (scribere engine or an instance), then read AGENTS.md and the relevant specs (docs/build.md, docs/ops.md, docs/templating.md, docs/queries.md) before proposing changes. Treat AGENTS.md as mandatory. It is the entry point for any work in this repo and must be followed throughout.

For first‑time setup, use npx --yes github:jhlagado/scribere#main in a fresh folder. This creates the repo, installs Scribere, and copies /example/ into /content/. Do not re‑copy content if /content/ already exists.

For updates, run npm run update. This refreshes the Scribere dependency, syncs the required npm scripts, and ensures .gitignore and AGENTS.md match current defaults.

For local work, use npm start to build, serve, and watch. This writes the lint report to temp/lint-report.json without blocking the dev server. Use npm run build for a clean one‑shot build, and npm run lint -- --all when a full lint sweep is needed.

If a site switches to a custom domain, update content/site.json with:

npm run domain

Instance ownership is strict: templates, assets, client JS, content, and instance queries live under /content/ in the instance repo. Engine scripts stay at the repo root. Avoid moving templates into the engine unless explicitly requested.

Output checks

Ensure builds read from /content/ (falling back to /example/ only in the engine repo), and that no instance files are overwritten by updates. Verify that template changes remain pure HTML and that queries remain JSON‑only.