AI Agent Skills directory

Search the open Agent Skills ecosystem from inside a session and install what matches the task.

Pressure-test a plan with relentless questions until the holes are visible.

Ship production UI with a point of view instead of the default purple-gradient look.

Drive a real browser with a session that survives across commands.

Apply ranked React and Next.js performance rules instead of vague optimize-this advice.

Audit UI code against Vercel Web Interface Guidelines for design and accessibility.

Turn a repeatable workflow into a valid SKILL.md with frontmatter, triggers, and references.

Read, split, merge, fill, and create PDF files, including scanned documents.

Create and edit PowerPoint decks from outlines, articles, or existing .pptx files.

Build, clean, and analyze spreadsheets without dumping raw CSV into the chat.

Design and scaffold Model Context Protocol servers with real tool schemas.

Build a tight feedback loop before you guess. Reproduce, isolate, then patch.

Design deep modules: small interfaces hiding a lot of behavior, with tests at the seams.

Capture terms, invariants, and decisions before the code hardens the wrong nouns.

Run source-backed research in the background and return a cited brief, not a vibe summary.

Strip AI writing tells so the text sounds like a specific person, not a template.

A stateful tutor that remembers where you stopped and checks understanding before moving on.

Persist decisions, preferences, and open threads so the next session does not start blank.

Stop Vue 3 hallucination: current APIs, composition patterns, and the mistakes models still repeat.

Technical and on-page SEO: titles, canonicals, indexation, and Core Web Vitals clues.

Write page copy that is specific, active, and aimed at one action, not a slogan wall.