Model Quantization

Build high-quality MCP (Model Context Protocol) servers that let LLMs interact with external services through well-designed tools. Use when creating MCP servers to integrate APIs or services in Python (FastMCP) or Node/TypeScript (MCP SDK).

Assesses and rates quality 0-10 across multiple dimensions (correctness, maintainability, security, performance, testability, simplicity) with pros/cons analysis. Compares against project conventions and prior decisions from memory. Produces structured evaluation reports with actionable improvement suggestions. Use when evaluating code, designs, architectures, or comparing alternative approaches.

Manage Git commits using conventional commit format with atomic staging. Always generate plain git commands before running them and offer to let the user run them manually.

Runs a second-pass cleanup over AI-written code using the repo's style guide in style.md. Prefers parallel subagents to simplify recently modified files without changing behavior. Use when the user says "deslop", "clean this up", "make this less AI", "apply my style guide", "second pass", or asks to simplify generated code after implementation.

Plan an animated product video end-to-end: pick the video type, define the animation style and pacing, write a beat-by-beat shot list, and produce ready-to-run keyframe prompts. Use whenever the user wants a product video, product animation, launch video, promo or ad video, animated product reveal, 360 spin, feature showcase, or 'a short video for my product' — even if they only say 'I need a video for the launch' or 'make my product look good in motion'. Also use to storyboard a video another tool will render. Produces a Product Video Production Spec: concept, beat sheet with timings and transitions, keyframe prompt list for fal-ai/nano-banana-pro, and audio/CTA direction.

Run deterministic accessibility audits and regression gates on web projects, especially large sites with many routes and shared templates. Combines axe-core and optional Lighthouse with template-aware sampling, stable finding fingerprints, accepted baselines, standards evidence mapping (WCAG 2.1 AA default, WCAG 2.2 AA, EN 301 549), manual check guidance, and structured reporting. Output is configurable: markdown report only, markdown plus machine-readable JSON, or markdown plus issue tracker integration. Use this skill whenever the user mentions "accessibility audit", "a11y audit", "WCAG audit", "accessibility check", "compliance scan", "accessibility baseline", "new accessibility regressions", or asks to check a web project for accessibility issues. Also trigger when the user wants evidence for WCAG conformance review or mapping to a specific standard (WCAG 2.2, EN 301 549 / European Accessibility Act, CAN-ASC-6.2, ADA/AODA).