reference-pack

Explains the gbrain skillpack contract and its ten scored doctor dimensions.

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

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Authors of third-party gbrain skillpacks need to know exactly what artifacts a pack must ship and how the doctor scores it before publishing. This skill answers "what is a skillpack" questions by describing the required files, the ten scoring dimensions, and the tier eligibility rules. ## Core Features & Use Cases - Contract Reference: Documents the four artifacts every pack ships: skillpack.json, SKILL.md with frontmatter and triggers, runbooks/bootstrap.md, and CHANGELOG.md. - Doctor Scoring Explained: Breaks down the 10 binary dimensions split into 5 core gates (manifest_valid, skills_have_skill_md, routing_evals_present, skills_have_unique_triggers, changelog_present_and_current) and 5 quality badges. - Tier Eligibility Guidance: Clarifies endorsed, community, experimental, and blocked tiers plus the doctor and pack commands publishers run before submitting. - Use Case: A developer asks their agent "how do third-party skillpacks work" and receives a grounded explanation of the manifest, routing evals, and publish checklist, plus pointers to gbrain skillpack doctor for live scoring. ## Quick Start Ask your agent what a gbrain skillpack is and how the doctor scores it.

Frequently Asked Questions about reference-pack

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

FAQPage Schema
What is a gbrain skillpack?

A gbrain skillpack is a third-party bundle shipping a skillpack.json manifest, one or more skills with SKILL.md files, a bootstrap runbook, and a changelog. The gbrain doctor scores it against ten binary dimensions before publication.

How do I publish a third-party gbrain skillpack?

Run gbrain skillpack doctor to check all scoring dimensions, fix failures with the paste-ready suggestions or --fix --yes, then run gbrain skillpack pack to emit a versioned tarball with a deterministic SHA-256 for the registry PR.

What files must a skillpack contain to pass the doctor?

The five core dimensions require a schema-valid skillpack.json, SKILL.md with name, description, and triggers for every skill, a routing-eval.jsonl with at least five intents per skill, unique triggers across the pack, and a CHANGELOG entry matching the manifest version.

What is the difference between endorsed, community, and experimental tiers?

Endorsed requires all ten dimensions plus a registry endorsement overlay. Community requires all five core dimensions plus at least three of five quality badges. Experimental passes all core dimensions but fewer than three badges; any core failure blocks the pack.

Why does the changelog dimension fail on my skillpack?

The changelog_present_and_current dimension fails when CHANGELOG.md lacks a ## [<version>] entry matching the version declared in skillpack.json. Add a Keep-a-Changelog entry for the current version to pass.