Skip to content
Skill Authoring
Skill

/skill-creator

Trigger: new skills, agent instructions, documenting AI usage patterns. Create LLM-first skills with valid frontmatter.

BOOST
From plugin
gentle-ai
7.8k18 skills11 agents2 commands
Install
$ npx -y skills add Gentleman-Programming/gentle-ai --skill skill-creator --agent claude-code

How it fires

How this skill gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
  • You can call itInvoke it directly when you want it.
  • Slash command/skill-creator

Context preview

The summary Claude sees to decide when to auto-load this skill.

Trigger: new skills, agent instructions, documenting AI usage patterns. Create LLM-first skills with valid frontmatter.

SKILL.md

skill-creator.SKILL.md
name: skill-creator
description: "Trigger: new skills, agent instructions, documenting AI usage patterns. Create LLM-first skills with valid frontmatter."
license: Apache-2.0
metadata:
  author: gentleman-programming
  version: "1.0"

Activation Contract

Create a skill when:

  • A pattern is used repeatedly and AI needs guidance
  • Project-specific conventions differ from generic best practices
  • Complex workflows need step-by-step instructions
  • Decision trees help AI choose the right approach

Do not create a skill when the pattern is trivial, one-off, or better served by normal documentation.

Hard Rules

  • When working in this repo, first follow `docs/skill-style-guide.md` as the normative source before creating or updating skills.
  • For installed global skills, use `references/skill-style-guide.md` as the bundled local copy of that guide when `docs/skill-style-guide.md` is unavailable.
  • If neither guide is available, use the compact inline rules below.
  • A skill is a runtime instruction contract for an LLM, not human documentation.
  • Do not add a `Keywords` section; preserve essential trigger words in `description`.
  • References must point to local files.
  • Keep the skill body concise: target 180–450 tokens, recommended max 700, hard max 1000.

Decision Gates

| Need | Action | |------|--------| | Code templates, schemas, fixtures, generated examples | Put them in `assets/` | | Conceptual detail, edge cases, existing docs | Put local links in `references/` | | Long explanation in `SKILL.md` | Move it to a supporting file | | Multiple meaningful paths | Add a compact decision table |

Execution Steps

1. Check whether `docs/skill-style-guide.md` exists; if it does, apply it before the bundled local copy or inline fallback rules. 2. If the repo guide is unavailable, read `references/skill-style-guide.md` and apply it before the inline fallback rules. 3. Confirm the skill does not already exist and the pattern is reusable. 4. Create or update `skills/{skill-name}/SKILL.md` using this required structure:

skills/{skill-name}/
├── SKILL.md              # Required - main skill file
├── assets/               # Optional - templates, schemas, examples
│   ├── template.py
│   └── schema.json
└── references/           # Optional - links to local docs
    └── docs.md           # Points to docs/developer-guide/*.mdx

5. Use this frontmatter shape:

---
name: {skill-name}
description: "Trigger: {essential trigger words users or agents will say}. {What this skill does}."
license: Apache-2.0
metadata:
  author: "{your-github-username}"
  version: "1.0"
---

6. Write sections in this order: Activation Contract, Hard Rules, Decision Gates, Execution Steps, Output Contract, References. 7. Register the skill in `AGENTS.md` when it is a project skill.

Inline Fallback Rules

  • `description` MUST be one physical line, quoted, YAML-safe, and include essential trigger words first.
  • `description` SHOULD be <=160 chars and MUST be <=250 chars.
  • Frontmatter MUST include `name`, `description`, `license`, `metadata.author`, and `metadata.version`.
  • Use imperative instructions, not tutorials or background prose.
  • Put supporting material in `assets/` or `references/`, not the main skill body.

Good:

description: "Trigger: Jira task, ticket, issue, task creation. Create Jira tasks in the team format."

Bad:

description: >
  Create Jira tasks in the team format.
  Trigger: Jira task, ticket, issue, or task creation.
Keywords: jira, task

Output Contract

Return:

  • Files created or modified.
  • Whether the repo style guide or inline fallback rules were used.
  • Any AGENTS.md registration change.
  • Any supporting files added under `assets/` or `references/`.

References

  • `docs/skill-style-guide.md` — normative LLM-first skill style guide for this repo.
  • `references/skill-style-guide.md` — bundled local copy for installed global skills when the repo doc is unavailable.
Read more
Ships withgentle-ai

Gentle-AI configures the AI coding agents you already use: Claude Code, Cursor, OpenCode, Codex, Pi, and more. Choose persistent memory, Organic-Driven Development, curated skills, MCP servers, personas, and optional bounded review. Open source, no agent lock-in.

Get the whole plugin

Other skills on gentle-ai.