schema-author

Evolves GBrain schema packs by adding page types, link verbs, and backfilling page.type on existing pages.

Updated Aug 17, 2026
One-click install
npx skills add https://github.com/AMC-JTC/gbrain-1 --skill schema-author-amc-jtc
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: schema-author
Source: https://github.com/AMC-JTC/gbrain-1/tree/main/plugin/skills/schema-author
Command: npx skills add https://github.com/AMC-JTC/gbrain-1 --skill schema-author-amc-jtc

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Your knowledge brain accumulates untyped pages that search and expert-routing queries cannot reach. This Skill lets you author and evolve the schema pack itself — adding page types, aliases, prefixes, and link verbs — then backfill existing pages so the new structure actually takes effect. ## Core Features & Use Cases - Schema mutation: Add, update, or remove page types, aliases, prefixes, and link types via CLI or batched MCP operations, with atomic writes and audit logging. - Corpus-driven proposals: Run gbrain schema detect and gbrain schema suggest to cluster untyped pages by path and propose candidate types with confidence scores. - Safe backfill: Dry-run gbrain schema sync to preview per-prefix would_apply counts, then apply chunked, idempotent updates to backfill pages.type. - Use Case: You imported 4,000 meeting notes that are invisible to typed queries. Use this Skill to add a meeting type with the meetings/ prefix, lint the pack, sync to backfill, and verify coverage rises above 95%. ## Quick Start Ask the agent to add a researcher page type with the people/researchers/ prefix to my schema pack and sync existing pages to backfill the new type.

Frequently Asked Questions about schema-author

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I add a new page type to my GBrain schema pack?

Run `gbrain schema add-type <name> --primitive entity --prefix <path>/` after forking the bundled pack with `gbrain schema fork gbrain-base mine`. Then run `gbrain schema sync --apply` to backfill matching existing pages.

How do I find untyped pages in my knowledge base?

Run `gbrain schema stats --json` for coverage percentages and `gbrain schema review-orphans --json` for a drilldown of untyped pages. Shared path prefixes among orphans indicate candidates for a new type.

Can I mutate the bundled gbrain-base schema pack directly?

No. Bundled packs like gbrain-base and gbrain-recommended are read-only and mutations fail with PACK_READONLY. Fork first with `gbrain schema fork gbrain-base mine`, then activate it with `gbrain schema use mine`.

Why does schema sync need a dry run before applying?

The dry run shows per-prefix `would_apply` counts and sample slugs so you can verify a prefix matches the intended pages. A mis-declared prefix could match tens of thousands of pages, so previewing first prevents costly mistakes.

What does the STILL_REFERENCED error mean when removing a type?

It means another type's aliases, enrichable_types, link_types, or frontmatter_links still reference the type you are removing. The error names every reference; remove those references first rather than forcing the deletion.

When should I not add a new schema type?

Avoid adding types for one-time imports or directories with fewer than about 20 pages, since pack types are permanent decisions. For filing a single note, use the brain-taxonomist skill instead of changing the schema.