Creating Agent Skills

Create or improve Agent Skills (the shared SKILL.md format read by Claude Code and OpenAI Codex CLI — SKILL.md + optional references/scripts/assets). Two playbooks: build a skill from scratch or from an existing doc; or review an existing skill through parallel specialist lenses (prose quality, agent neutrality, structure, scripts, orchestration) and propose fixes. Takes a free-form request — "I want a skill that…", "turn this runbook into a skill", "audit this skill", "it never fires", "split this into sub-agents" — optionally followed by a skill name or absolute path. Also triggers on Codex

5 starsAI, LLMs & Agents Plugin marketplace

Install

Claude Code plugin
/plugin marketplace add tomada1114/tomada-claude-plugins

The repo is a Claude Code plugin marketplace. Run /plugin afterwards to pick the plugin that includes this skill.

skills CLI (Claude Code, Cursor, Codex and more)
npx skills add tomada1114/tomada-claude-plugins --skill creating-agent-skills
Manual install
git clone --depth 1 https://github.com/tomada1114/tomada-claude-plugins.git /tmp/tomada-claude-plugins
mkdir -p ~/.claude/skills
cp -r /tmp/tomada-claude-plugins/tomada-kit/skills/creating-agent-skills ~/.claude/skills/creating-agent-skills

Personal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.

From the SKILL.md

# creating-agent-skills

Workshop for building and maintaining Agent Skills — the shared `SKILL.md` format both Claude Code and Codex CLI read natively (see [references/agent-neutral-authoring.md](references/agent-neutral-authoring.md) for what's common vs. per-host). Most guidance here is host-agnostic; the few Claude Code-only mechanics (`Task`, `AskUserQuestion`, plugin `${CLAUDE_PLUGIN_ROOT}`, etc.) are called out explicitly where they appear.  

`$ARGUMENTS` is **free-form text**, in any language. There is no mode keyword to parse. Read the user's intent, optionally pick out a skill name or path token, then route to [Building a skill](#building-a-skill) or [Improving a skill](#improving-a-skill). If the request mixes goals (e.g. "audit it, and fix whatever is broken"), chain playbooks in the order implied by the request.

## Contract

**Input:** free-form natural-language description of the task, optionally including a skill name or absolute path.

**Outputs depend on the inferred task:**
- Building → directory at `~/.claude/skills/<name>/` (or `<cwd>/.claude/skills/<name>/` with `--scope project`). Run `dual-platform-skills` afterward to bridge it onto Codex. 
- Improving → `${AGENT_SKILL_STATE_DIR:-$HOME/.local/state/agent-skills}/skills-audit/<skill-name>/` holding `report.md`, `lens-<name>.md` per lens, `recommendations.md`. Edits to the skill itself only after the user approves.

**Bundled scripts** (invoke through `${CLAUDE_SKILL_DIR}`, never a hardcoded path):
Read the full skill on GitHub

Details

Skill name
creating-agent-skills
Path
tomada-kit/skills/creating-agent-skills/SKILL.md
Repo updated
12 November 2026

FAQ

What does the Creating Agent Skills skill do?

Create or improve Agent Skills (the shared SKILL.md format read by Claude Code and OpenAI Codex CLI — SKILL.md + optional references/scripts/assets). Two playbooks: build a skill from scratch or from an existing doc; or review an existing skill through parallel specialist lenses (prose quality, agent neutrality, structure, scripts, orchestration) and propose fixes. Takes a free-form request — "I want a skill that…", "turn this runbook into a skill", "audit this skill", "it never fires", "split this into sub-agents" — optionally followed by a skill name or absolute path. Also triggers on Codex

How do I install the Creating Agent Skills skill?

Claude Code plugin: /plugin marketplace add tomada1114/tomada-claude-plugins. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.

Is Creating Agent Skills popular?

It lives in tomada1114/tomada-claude-plugins, which has 5 GitHub stars and 0 forks.

Related skills

Auditing Prompt Docs

5

tomada1114/tomada-claude-plugins

Audit and fix prompt-bearing documents — SKILL.md, CLAUDE.md, AGENTS.md, agent and command definitions, reference files, whole repositories — against Anthropic's published prompting guidance for current Claude models, proposing edits for approval before writing anything. Also answers how to prompt a specific model: holds the per- model guidance for Claude Fable 5.2, Claude Opus 5.5, Claude Sonnet 5.5, and Claude Haiku 5.5 alongside the cross-model practices. Use when reviewing or slimming an instruction file, when instructions are ignored or fire too often, when migrating a prompt to a newer m

AI, LLMs & Agents

Auditing Repositories

5

tomada1114/tomada-claude-plugins

Audit the repository you are in, often a template or an app cut from one, for technical debt and room to improve in four areas: architecture and design patterns; tests; file structure and code quality; and tooling, CI, dependencies and the agent harness. Design a safe, test-first fix for every evidenced finding, and file the findings as GitHub issues that follow the repository's own issue conventions (triaging-issues, issue forms, past issues). A finding inherited from the template is also filed in the template repository. Cheap read-only sub-agents collect the evidence and high-effort analyst

Testing & QA

Authoring Goal Prompts

5

tomada1114/tomada-claude-plugins

Author a self-contained prompt for Claude Code's /goal command, run unattended in a separate session — a lead-to-staff handoff where this session researches, designs, and decides with full context, and a fresh executor session (Opus 5.5 or Fable 5.2, with none of that context) implements. Drafts a goal with measurable done-criteria, transcript-verifiable checks, scope + anti-cheat constraints, and stop rules; packages design intent, code examples, research findings, checklists, and pre-answered decisions as support files in a per-goal state directory whenever the executor would otherwise re-de

AI, LLMs & Agents

Debate Claude

5

Magnus-Gille/claude-skills

Run a two-round adversarial review from Codex with Claude as the independent reviewer for a consequential decision or substantive draft when cross-model critique is warranted.

AI, LLMs & Agents

Configuring Project Permissions

5

tomada1114/tomada-claude-plugins

Set up a repository's personal, uncommitted agent permissions so development runs without prompts: Claude Code's gitignored .claude/settings.local.json and Codex CLI's gitignored .codex/rules/local.rules, each in its native format. Generic rules (git, gh, coreutils, hook-bypass and force-push denies) go to the user-level files; the project file keeps only what names the repository — recipes, scripts, dependency policy, protected files — allow by default, ask only for commands that open the app, write to GitHub, release, or change dependencies. Detects the toolchain, drops rules a global rule a

AI, LLMs & Agents

Delegate

5

Magnus-Gille/claude-skills

Delegate a bounded one-shot task to Hugin's orchestrator-v1 broker — cheaper / smaller runtimes via the `hugin-mcp` MCP. Use for summarize / extract / classify / draft / rewrite / short reason work where you'd rather not spend Claude tokens. Submits, polls to completion, surfaces the result, and prompts the user to rate it.

AI, LLMs & Agents

Browse the most popular Claude skills, all ai, llms & agents skills, or write your own with the skills generator.