How do I install the Arc Creating Skill skill?

Run npx skillmds@latest add andysolomon/arc-creating-skill in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Arc Creating Skill safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Arc Creating Skill?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Arc Creating Skill free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Arc Creating Skill?

andysolomon (@andysolomon) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Arc Creating Skill by andysolomon — Agent Skill

Name: Arc Creating Skill
Author: andysolomon

Arc Creating Skill

Creates or revises agent skills in agentskills.io `SKILL.md` format. Use when the user asks to create a new skill, turn notes into a skill, improve a skill, split/consolidate skills, apply skill-writing best practices, or make a skill predictable and concise.

andysolomon c9bb192 4 files · 8.8 KB Updated Sep 17, 2026

File contents

Creating Skills

Create or revise a skill so it predictably changes agent behavior without carrying unnecessary context.

Steps

Classify invocation. Decide whether the skill is model-invoked or user-invoked. Keep a trigger-rich description only when the agent or another skill must discover it; otherwise set disable-model-invocation: true and make the description human-facing. Completion: the invocation choice and its reason are stated.
Draft the behavior spine. Write the smallest ordered steps that make the skill work. End every step with a checkable completion criterion. Completion: each step says how the agent knows that step is done.
Disclose reference. Keep only always-needed steps in SKILL.md; move branch-specific templates, rubrics, examples, command recipes, and long checklists to linked sibling markdown files. Completion: SKILL.md is short enough to scan and every pointer says when to read its target.
Prune for predictability. Apply GREAT_SKILL_PRINCIPLES.md: remove duplication, sediment, no-ops, vague encouragement, and weak trigger prose. Prefer one strong leading word over repeated explanation. Completion: every remaining line changes invocation or execution behavior.
Add eval hooks. If the repo uses arc-skill-eval, add or update evals/evals.json with at least one positive path and one adjacent negative. Completion: the skill has runnable coverage or an explicit reason coverage is deferred.
Report the contract. Summarize created/changed files, invocation mode, disclosed references, eval coverage, and any assumptions the user should verify. Completion: the user can run or install the skill from the paths you provide.

Output contract

Produce file changes, not just advice, when the user asks to create or revise a skill. If information is missing, make the smallest safe assumption and call it out rather than blocking, unless the invocation mode or target skill name is ambiguous.

---name: arc-creating-skilldescription: Creates or revises agent skills in agentskills.io `SKILL.md` format. Use when the user asks to create a new skill, turn notes into a skill, improve a skill, split/consolidate skills, apply skill-writing best practices, or make a skill predictable and concise.---# Creating SkillsCreate or revise a skill so it predictably changes agent behavior without carrying unnecessary context.## Steps1. **Classify invocation.** Decide whether the skill is model-invoked or user-invoked. Keep a trigger-rich `description` only when the agent or another skill must discover it; otherwise set `disable-model-invocation: true` and make the description human-facing. Completion: the invocation choice and its reason are stated.2. **Draft the behavior spine.** Write the smallest ordered steps that make the skill work. End every step with a checkable completion criterion. Completion: each step says how the agent knows that step is done.3. **Disclose reference.** Keep only always-needed steps in `SKILL.md`; move branch-specific templates, rubrics, examples, command recipes, and long checklists to linked sibling markdown files. Completion: `SKILL.md` is short enough to scan and every pointer says when to read its target.4. **Prune for predictability.** Apply [`GREAT_SKILL_PRINCIPLES.md`](GREAT_SKILL_PRINCIPLES.md): remove duplication, sediment, no-ops, vague encouragement, and weak trigger prose. Prefer one strong leading word over repeated explanation. Completion: every remaining line changes invocation or execution behavior.5. **Add eval hooks.** If the repo uses `arc-skill-eval`, add or update `evals/evals.json` with at least one positive path and one adjacent negative. Completion: the skill has runnable coverage or an explicit reason coverage is deferred.6. **Report the contract.** Summarize created/changed files, invocation mode, disclosed references, eval coverage, and any assumptions the user should verify. Completion: the user can run or install the skill from the paths you provide.## Output contractProduce file changes, not just advice, when the user asks to create or revise a skill. If information is missing, make the smallest safe assumption and call it out rather than blocking, unless the invocation mode or target skill name is ambiguous.

Frequently asked questions