Help us improve
Share bugs, ideas, or general feedback.
From virtual-team
Defines structure, size budget, description format, and review checklist for creating or modifying skills in the plugin.
npx claudepluginhub ovargas/virtual-team --plugin virtual-teamHow this skill is triggered — by the user, by Claude, or both
Slash command
/virtual-team:skill-authoringThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
SKILL.md files must stay **under 100 lines**. If content exceeds this, split into reference files in a `references/` subdirectory.
Generates validated SKILL.md files for Claude Code skills with YAML frontmatter, naming conventions, tool minimalism, structure constraints, and best practices.
Guides technical evaluation of code review feedback: read fully, restate for understanding, verify against codebase, respond with reasoning or pushback before implementing.
Share bugs, ideas, or general feedback.
SKILL.md files must stay under 100 lines. If content exceeds this, split into reference files in a references/ subdirectory.
disable-model-invocation: true in frontmatter — they are data, not behavioral promptsRequired fields (enforced by tests/validate-frontmatter.test.ts):
| Field | Rule |
|---|---|
name | Matches directory name, kebab-case |
description | See Description Format below |
Optional fields:
| Field | When to use |
|---|---|
model | Override the default model (e.g., model: sonnet for mechanical tasks) |
loaded_when | Document which commands load this skill |
disable-model-invocation | Set true on reference files — marks them as data-only |
Good: Use when implementing any feature or bugfix, before writing implementation code — enforces red-green-refactor cycle with configurable strictness
Bad: TDD skill for testing (no trigger, no enforcement description)
skills/<name>/
SKILL.md — Main file (always loaded, under 100 lines)
references/ — Optional, lazy-loaded sub-topic files
example.md — Data tables, examples, checklists
skills/git-practices/SKILL.md.claude/skills/... or relative ../ pathsSKILL.md contains behavioral rules — the instructions the AI must follow when the skill is active. Keep it focused: conditions, rules, verification steps.
Reference files contain supporting data — comparison tables, example lists, rationalization catalogs, detailed checklists. They are read only when the specific sub-topic is relevant.
This split is how you stay under 100 lines without losing depth. See skills/test-driven-development/ for the model: SKILL.md at 98 lines, with references/test-quality.md and references/rationalizations.md loaded on demand.
Scripts are for deterministic operations only: file manipulation, formatting, validation, template generation. Never use scripts for AI judgment calls.
scripts/ at the repo rootBefore merging a new or modified skill, verify:
name and descriptiondisable-model-invocation: true in frontmatternpm test passes (frontmatter validation, file reference validation)