AI Agent Skills
Hand-picked Skills for Claude Code and other AI agents, most with a real-run demo and all with source, license and risk notes. Anything we have not tried is clearly marked.
Frontend Design
Guidance for building distinctive, intentional UI instead of templated defaults: plan the design first, then build.
Systematic Debugging
Find the root cause before you fix anything: a four-phase process of evidence, patterns, one hypothesis at a time and a tested fix, with a stop rule after three failed attempts.
Image Prompt Writing
Turn a vague idea into a clear prompt for an AI image generator: subject, setting, composition, light, style and color in a sensible order, with a short version, a detailed version and variations.
X Thread Writer
Write posts for X that say one thing clearly: a single sharp post, a thread, a quote post or a reply, with a standalone first post, one idea per post and an edit pass that cuts the filler.
E-commerce Unit Economics
Find out what one order really earns after product cost, shipping, fees, returns and ads, then get break-even and target ROAS or CPA, a price for a target margin, and a check on how deep a discount can go.
Conversion Rate Optimization (CRO)
Analyse a marketing page or form and get ranked recommendations: value proposition, headline, CTA, trust signals, objections and friction, with test ideas and copy alternatives.
Conversion Copywriting
Write and rewrite marketing copy for homepages, landing, pricing and feature pages: clear headlines, strong CTAs, page structure, and a strict list of AI tells to avoid.
Paper Reading
Read one research paper critically in three passes: what it claims, how it tests the claim, whether the evidence supports it, and what you can take from it, with quotes for the key claims and "not stated" where the text is silent.
MarkItDown: Files to Markdown
Convert PDF, Word, PowerPoint, Excel, HTML and other files to clean Markdown with Microsoft MarkItDown, for search, analysis and feeding documents to an AI, with safe local defaults.
Handoff
Compacts the current conversation into a handoff document so a fresh agent session can pick the work up.
TDD
Test-first development one thin slice at a time: write a failing test, make it pass, and test only at seams you agreed on.
Prototype
Builds a throwaway prototype to answer one design question: a clickable HTML model for logic, or several UI variations to compare.
Grill with Docs
Interview you about a plan or design and write down the decisions as ADRs and a glossary along the way.
Grill Me
Start a relentless interview that sharpens a plan or design, one question at a time.
SEO Audit
Produce evidence-based technical, on-page, and multilingual SEO audits with prioritized fixes.
Internal Comms
Write internal communications in fixed company formats: 3P updates, company newsletters, FAQ answers and general announcements.
Theme Factory
Apply one of 10 ready-made color and font themes to slides, docs, reports or landing pages, or generate a new theme on the fly.
Chrome DevTools Agent
Control and inspect a live Chrome browser through the Chrome DevTools MCP: navigate, click, fill forms, take snapshots and screenshots, read the console and network, and profile performance.
Domain Name Brainstormer
Generates domain name ideas for your project across several extensions, and checks availability when your agent can look it up.
MCP Builder
A guide for building high-quality MCP servers in TypeScript or Python: research, design, implementation, testing and evaluations.
Google SEO Audit
Audit a page or site for Google Search from the HTML, robots.txt and Search Console data you give it: crawl and indexing blockers, titles, content quality, structured data and page experience, as a prioritized fix list.
Receiving Code Review
Handle review feedback with technical rigour: understand it, check it against the code, push back when it is wrong, clarify before acting and implement one item at a time.
Discernment Nudge
After a substantive answer you may act on, adds 2-3 specific follow-up questions that help you check facts, probe the reasoning and notice missing context.
Doublecheck (AI output verification)
Check AI-written text: extract every verifiable claim, look for sources you can open yourself, and flag likely hallucinations in a structured report.