Added Nov 4, 2026
agent-skills
mblode's Skill set for coding agents, covering UI audits, typography, docs, PR review and releases.
- Ecosystem
- Agent Skills
- Skill category
- Coding
- Creator / organization
- mblode(GitHub 用户)
What it is
agent-skills is a set of coding-agent Skills by Matthew Blode (GitHub user mblode). The README pitches them as a way to avoid shipping sloppy AI output, covering UI audits, typography, docs, PR review and releases, grouped into architecture, design, writing and more. We sampled a few: agent-ready uses reports from scorers such as Is Agentic, or server logs, to make public sites and docs easier for AI agents to read; typography-audit checks font loading, type scales and punctuation with 78 rules; chat-history recovers past decisions and fixes from local Claude Code, Codex, Grok and other chat logs; autoship runs an npm package release. We didn't read the other items one by one.
Highlights:
- One command installs it for both Codex and Claude Code
- each Skill states what it covers and what it leaves to others
- the design group includes UI audits, animation and deck building.
How to use:
- First run npx skills add mblode/agent-skills -g --agent codex claude-code -y.
- Then ask for a Skill by name, for example have typography-audit check your project's fonts.
Who it's for:
Developers building websites and front-end products with Codex or Claude Code who care about finish.
Before you start:
- You need Codex or Claude Code, and Node and npm for the install command.
- agent-ready's check script needs Bash and curl and requests your site.
- chat-history reads the AI chat logs on your computer.
- pr-babysitter and autoship act on your GitHub PRs and npm releases.
- If you use these features, weigh the safety risks yourself.
- AI Inspiration World is not responsible for any consequences.
- Your agent's own costs are separate.
- MIT license.
Based on the project's public documentation. Not tested by AI Inspiration World.