Skills

Templates and guidelines for writing company-standard internal communications across multiple formats.

Covers seven communication types: 3P updates, company newsletters, FAQ responses, status reports, leadership updates, project updates, and incident reports
Each format has a dedicated guideline file in the examples/ directory with specific instructions for tone, structure, and content
Helps maintain consistency across internal communications by enforcing company-preferred formats and styles
Falls back to general communication guidelines for formats not explicitly covered by specialized templates

Insert instructions below

Distinctive, production-grade frontend interfaces that reject generic AI aesthetics.

Guides design thinking before coding: establish bold aesthetic direction (brutalist, maximalist, retro-futuristic, luxury, playful, etc.) with clear purpose, tone, and differentiation
Emphasizes typography, color theming, motion, spatial composition, and textural details as core design pillars; uses CSS variables, animations, and scroll-triggered effects
Generates working, functional code (HTML/CSS/JS, React, Vue) that matches implementation complexity to aesthetic vision: elaborate code for maximalist designs, restrained precision for minimalist ones
Explicitly avoids generic patterns: overused fonts (Inter, Roboto), clichéd color schemes (purple gradients), predictable layouts, and cookie-cutter component design

Toolkit for creating animated GIFs optimized for Slack with validation and animation utilities.

Provides GIFBuilder for frame assembly, validators to check Slack compliance, and easing functions for smooth motion (bounce, elastic, ease-in-out, etc.)
Supports emoji GIFs (128×128, under 3 seconds) and message GIFs (480×480) with configurable FPS, color palette, and optimization flags
Includes animation concepts (shake, pulse, bounce, spin, fade, slide, zoom, particle burst) with code examples and PIL primitives for drawing shapes
Built on PIL for frame creation and manipulation; handles user-uploaded images as direct content or style reference

Create, test, and iteratively improve AI agent skills with structured evaluation and benchmarking.

Guides you through the full skill development lifecycle: intent capture, drafting, test case creation, evaluation, and iteration based on user feedback
Runs parallel test cases with and without the skill to measure impact, capturing timing and token usage for quantitative comparison
Generates an interactive browser-based reviewer showing outputs, qualitative feedback, and benchmark metrics (pass rates, latency, token efficiency) across iterations
Includes description optimization to improve skill triggering accuracy by testing 20 realistic trigger/non-trigger queries and iterating the description through up to 5 rounds

Claude API integration for building LLM-powered applications across Python, TypeScript, Java, Go, Ruby, C#, and PHP.

Defaults to Claude Opus 4.6 with adaptive thinking and streaming; supports tool use, structured outputs, batches, and file uploads through a single /v1/messages endpoint
Language detection automatically routes you to the correct SDK documentation; includes decision trees for choosing between single API calls, workflows with tool use, and agentic loops
Tool runner (beta in most languages) handles automatic loop execution; manual loops available for fine-grained control over approval gates, logging, and conditional execution
Agent SDK (Python and TypeScript only) provides built-in file, web, and terminal tools with permissions, MCP support, and safety guardrails; Claude API is the right choice for custom agent tools

Comprehensive PDF processing with text extraction, merging, splitting, form filling, and OCR capabilities.

Supports core operations: merge/split PDFs, extract text and tables, rotate pages, add watermarks, encrypt/decrypt, and extract images
Includes Python libraries (pypdf, pdfplumber, reportlab) and command-line tools (qpdf, pdftotext, pdftk) with ready-to-use code examples
Handles scanned PDFs via OCR using pytesseract and pdf2image for searchable text extraction
Dedicated form-filling workflow documented in FORMS.md; advanced features and JavaScript alternatives covered in REFERENCE.md

Create, edit, read, and manipulate PowerPoint presentations with design guidance and quality assurance workflows.

Supports three primary workflows: reading/extracting text from existing .pptx files, editing presentations via template unpacking and repacking, and creating decks from scratch using PptxGenJS
Includes 10 curated color palettes and typography pairings designed to avoid generic AI aesthetics, plus layout patterns (two-column, icon rows, grids, half-bleed images) with specific spacing and contrast rules
Provides mandatory QA process using subagents for visual inspection, with automated checks for placeholder text, overlapping elements, text overflow, and low-contrast content
Converts slides to JPEG images for detailed inspection and includes tools for text extraction, thumbnail generation, and XML unpacking for advanced editing

Create, read, edit, and manipulate Word documents (.docx files) with full formatting control.

Create new .docx files from scratch using JavaScript with support for headings, tables, images, hyperlinks, footnotes, headers/footers, and table of contents
Edit existing documents by unpacking XML, making targeted changes (including tracked changes and comments), and repacking with validation
Extract text and analyze content using pandoc or raw XML access; convert legacy .doc files to .docx format
Handle professional formatting: page sizes (US Letter/A4), margins, multi-column layouts, tab stops, and styled lists without manual unicode characters

Create, edit, and analyze Excel spreadsheets with formulas, formatting, and error-free calculations.

Supports reading, creating, and modifying .xlsx, .xlsm, .csv, and .tsv files using pandas for data analysis and openpyxl for formulas and formatting
Automatically recalculates all formulas and scans for Excel errors (#REF!, #DIV/0!, #VALUE!, #N/A, #NAME?) using LibreOffice integration
Enforces industry-standard color coding and number formatting for financial models (blue for inputs, black for formulas, green for internal links, red for external links)
Includes comprehensive documentation requirements for hardcoded values and assumption-based formula construction to maintain dynamic, updateable spreadsheets

Native Python Playwright scripts for testing local web applications with server lifecycle management.

Includes with_server.py helper to automatically start and manage single or multiple servers before running automation scripts
Supports reconnaissance-then-action workflow: screenshot/inspect rendered DOM, identify selectors, then execute actions with discovered selectors
Emphasizes waiting for networkidle on dynamic apps before DOM inspection to avoid timing issues
Provides example scripts covering element discovery, static HTML automation, and console log capture

Comprehensive guide for building high-quality MCP servers that connect LLMs to external services.

Four-phase workflow covering research, implementation, testing, and evaluation with language-specific guidance for TypeScript (recommended) and Python
Emphasizes API coverage over workflow tools, clear tool naming with consistent prefixes, and actionable error messages for agent guidance
Includes best practices for input/output schemas, pagination, response formatting, and tool annotations (readOnly, destructive, idempotent, openWorld hints)
Provides reference documentation links for MCP protocol specification, SDK documentation, and language-specific implementation patterns with code examples
Evaluation framework requiring 10 independent, read-only, complex questions with verifiable answers in XML format to validate LLM effectiveness with your server

Create visually-driven design artifacts from philosophical aesthetic movements, expressed as pristine PDF or PNG artworks.

Generates custom design philosophies (aesthetic movements articulated through form, space, color, and composition) as markdown foundations
Expresses philosophies visually through museum-quality single or multi-page PDF/PNG compositions with minimal, integrated text
Embeds subtle conceptual references invisibly into abstract compositions, rewarding deeper viewing without announcing intent
Prioritizes expert-level craftsmanship appearance: meticulous spacing, refined typography, cohesive color palettes, and flawless technical execution suitable for gallery or publication display

Generative art creation using p5.js with seeded randomness and parametric exploration.

Creates algorithmic philosophies (computational aesthetic movements) expressed through interactive p5.js sketches with full parameter control
Supports seeded randomness for reproducible art variations, seed navigation (previous/next/random/jump), and real-time parameter adjustment via sliders and color pickers
Outputs self-contained HTML artifacts with embedded p5.js code, sidebar controls, and download functionality—no external files or setup required
Emphasizes craftsmanship through controlled chaos: emergent behavior constrained by mathematical rules, noise fields, particle systems, and flow dynamics

Multi-component React artifacts for Claude with bundling, state management, and shadcn/ui component library.

Scaffolds React 18 + TypeScript projects with Vite, Tailwind CSS, and 40+ pre-installed shadcn/ui components via automated initialization script
Bundles complete artifacts into single self-contained HTML files with inlined assets, ready for Claude artifact display
Includes design guidance to avoid common «AI slop» patterns like excessive centering, purple gradients, and uniform rounded corners
Supports complex workflows requiring routing, state management, and Radix UI dependencies out of the box

Structured workflow for collaboratively authoring documentation, proposals, specs, and similar content.

Guides users through three stages: Context Gathering (closing knowledge gaps), Refinement & Structure (building sections iteratively), and Reader Testing (validating the doc works for fresh readers)
Supports integration with shared documents, team channels, and templates to pull in context directly when connectors are available
Uses brainstorming and curation cycles for each section, with surgical edits and iterative refinement based on user feedback
Includes reader testing via sub-agents (when available) or manual testing instructions to catch blind spots before others read the document

Professional font and color themes for styling presentation decks, documents, and web artifacts.

Includes 10 pre-set themes (Ocean Depths, Sunset Boulevard, Forest Canopy, Modern Minimalist, Golden Hour, Arctic Frost, Desert Rose, Tech Innovation, Botanical Garden, Midnight Galaxy) each with cohesive color palettes and complementary font pairings
Supports custom theme generation on-the-fly when none of the preset themes fit your needs
Applies themes consistently across artifacts by reading theme specifications and enforcing proper contrast and readability
Workflow: display theme showcase, get user selection, apply chosen theme’s colors and fonts throughout the artifact

Apply Anthropic’s official brand colors and typography to visual artifacts and designs.

Includes a defined color palette with primary dark/light neutrals, mid and light grays, plus three accent colors (orange, blue, green)
Applies Poppins font to headings (24pt+) and Lora font to body text, with automatic fallbacks to Arial and Georgia
Smart color selection based on background context to preserve readability and visual hierarchy
Works with system-installed fonts; no additional installation required, though pre-installing Poppins and Lora is recommended for best results

Discover and install specialized agent skills from the open ecosystem when users need extended capabilities.

• Helps identify relevant skills by domain and task when users ask «how do I do X» or «find a skill for X»
• Integrates with the Skills CLI (npx skills find, npx skills add) to search, verify, and install packages from the skills.sh directory
• Recommends skills based on install count, source reputation, and GitHub stars to ensure quality before suggesting installation
• Presents skill options with install commands and links to skills.sh for user review and one-click installation