Creates branded, editorial-style diagrams as self-contained HTML and SVG, and can redraw Mermaid, draw.io or Excalidraw sources.

Use when: You need an architecture diagram, flowchart or sequence chart that looks designed, not auto-generated.

Converts books and document folders (PDF, EPUB, DOCX and more) into a structured agent skill with frameworks, principles and techniques.

Use when: You want your agent to answer from a specific technical book instead of guessing.

Creates, reads and edits Word documents and templates, including formatting, tables and tracked changes.

Use when: You need to produce or fix a .docx file from the terminal.

Extracts text and tables from PDFs, merges, splits and rotates pages, fills forms and creates new PDF files.

Use when: You need to read, combine or generate PDF files.

Builds, reads and edits PowerPoint decks, from pitch decks to extracting text from existing presentations.

Use when: You need a slide deck made or changed as a real .pptx file.

Opens, edits, cleans and creates spreadsheets (.xlsx, .csv, .tsv) with formulas and formatting.

Use when: Your main input or output is a spreadsheet file.

Guides distinctive visual design for new or existing UI: aesthetic direction, typography, and choices that avoid templated defaults.

Use when: You are building a web page or component and want it to look intentional, not generic.

Creates original static visual art and posters as PNG or PDF files, guided by a written design philosophy.

Use when: You want a poster or one-off visual made from a prompt.

Helps you create new skills, improve existing ones, and measure skill performance with evals and benchmarks.

Use when: You are writing or tuning your own SKILL.md and want to test that it triggers.

A guide for building high-quality MCP servers that let an LLM use external services through well-designed tools.

Use when: You are wrapping an API or service as an MCP server.

Tests local web apps with Playwright: checks frontend behavior, captures screenshots and reads browser logs.

Use when: You need the agent to open your local app and check that it works.

A structured workflow for co-writing docs, proposals and specs: gather context, refine through iteration, then check the draft.

Use when: You are drafting a spec or decision doc and want a guided process.

Enforces test-first development: write a failing test, then the implementation, for every feature or bugfix.

Use when: You are about to implement a feature or bugfix and want tests first.

A disciplined debugging process: investigate the root cause before proposing any fix to a bug or failing test.

Use when: Something is broken and you want to stop the agent guessing at fixes.

Explores intent, requirements and design with you before any implementation starts.

Use when: You have a feature idea and want it shaped before code is written.

Turns a spec or requirements into a clear multi-step implementation plan before anyone touches code.

Use when: You have a spec for a multi-step task and need a plan.

Executes an implementation plan by handing independent tasks to fresh subagents in the current session.

Use when: You have a plan with independent tasks and want parallel, reviewed execution.

Makes sure feature work runs in an isolated workspace, using native tools or a git worktree fallback.

Use when: You are starting feature work and want it isolated from your current branch.

Runs a review of finished work against the requirements before you merge.

Use when: You finished a major feature and want a review pass before merging.

Runs a relentless interview to sharpen a plan or design before you build it.

Use when: You have a plan that feels half-baked and want hard questions first.

Scans a codebase for deepening opportunities, shows them as a visual HTML report, then grills you through the one you pick.

Use when: Your codebase is getting messy and you want a prioritized refactor map.

A diagnosis loop for hard bugs and performance regressions, triggered when something is broken, failing or slow.

Use when: A bug or slowdown has resisted your first few attempts.

Behavior rules that reduce common LLM coding mistakes: avoid overcomplication, make surgical changes, state assumptions, define success checks.

Use when: You want the agent to write less, change less and verify more.

React and Next.js performance guidelines from Vercel Engineering, applied when writing or reviewing components and data fetching.

Use when: You are writing or reviewing React or Next.js code for speed.

Reviews UI code against the Web Interface Guidelines for accessibility, UX and design best practices.

Use when: You want an audit of your UI code against solid guidelines.

Deploys apps and websites to Vercel, including preview deployments, and returns the live link.

Use when: You want to push a project live or get a preview URL.

A searchable design-intelligence database of styles, palettes, font pairings and UX rules for web and mobile interfaces.

Use when: You are designing or reviewing an interface and need concrete style and UX choices.

The router for Remotion Agent Skills: best practices for creating, animating, captioning and rendering programmatic videos in React.

Use when: You are making a video with Remotion and want the agent to follow its patterns.

Security-focused review of PRs, commits and diffs, scaled to codebase size, with blast-radius counts and a markdown report.

Use when: You are reviewing a pull request and want a security-minded pass.

Runs a Semgrep security scan: detects languages, proposes rulesets for your approval, then runs them in batches.

Use when: You want a static security scan of a codebase with a plan you approve first.

Lets a coding agent drive a browser from the command line: automate pages, run tests and work with Playwright tests.

Use when: Your agent needs to click through a real browser to test or inspect a page.

Rewrites AI-sounding text to read like the writer, based on Wikipedia's Signs of AI writing, without changing the meaning.

Use when: A draft sounds machine-written and you want the tells removed.

Researches what people said about any topic in the last 30 days across Reddit, X, YouTube, Hacker News, GitHub and the web.

Use when: You need fresh, grounded research on a topic or person.

Postgres best practices from Supabase for schema design, column types and migrations, for Postgres running anywhere.

Use when: You are creating or changing tables, queries or migrations in Postgres.

Writes Sentry-style conventional commit messages with issue references for every commit request.

Use when: You are about to commit and want a clean, consistent message.

Reviews pull requests using Sentry engineering practices, covering security, performance, testing and design.

Use when: You are reviewing a pull request and want a consistent checklist.