Skill UI

This skill reviews code differences (diffs) to identify opportunities for improvement in clarity, reusability, efficiency, and overall quality. It employs multiple specialized sub-agents (e.g., Code Reuse, Efficiency, Quality) to audit changes. It can provide high-confidence, behavior-preserving fixes when requested, ensuring the code adheres to best practices without introducing regressions. Best used when refactoring, cleaning up, or auditing a specific code scope.

Formats Markdown into Slack-ready HTML or mrkdwn, previews the Slack-style rendering, copies rich HTML to the clipboard, and optionally delivers announcements via webhooks for messages and notifications.

Refactors large SwiftUI views into small, stable components by enforcing explicit dependency injection, keeping business logic in services/models, and avoiding unnecessary view models so the UI tree stays predictable and maintainable.

Guides analysts through uncovering what customers think, feel, and say by analyzing interviews, surveys, support tickets, and community data or by sourcing fresh insights online so positioning, product, and copy stay grounded in real needs.

Assess product direction before building by validating the ‘why’, running diagnostics, and stress-testing decisions across diagnostics, founder reviews, user journey audits, and feature prioritization to ensure actionable recommendations.

Guides researchers through ICML-style rebuttal by parsing external reviews, enforcing coverage, grounding responses, respecting venue limits, and managing follow-up rounds for safe reviewer replies.

Provides REST API access to PubMed for building advanced Boolean/MeSH queries, filtering by publication details, and programmatically fetching citations through E-utilities, making it ideal for systematic reviews and automated literature workflows.

Applies two-axis scoring, mandatory devil’s advocate reasoning, and persistent history to keep post-task self-reviews honest, resistant to inflation, and suitable for Claude Code sessions.

Context Mode Ops orchestrates dozens of parallel subagents to manage GitHub issues, PR reviews, releases, marketing notices, and verification gates, ensuring every claim is verified, tests drive implementation, and grill interviews guard releases.

Generates three title tag and meta description variants per page, each following a distinct CTR angle, and adds native SERP previews plus bespoke Open Graph and Twitter Card copy for social sharing.

Generate CodeTour .tour files tailored to personas like new joiners, reviewers, architects, RCA investigators, and security auditors, using real files and line anchors for onboarding, PR reviews, architecture tours, or “explain how this works” guides.

Generates structured peer-review quality critiques of academic papers, covering summaries, methodology scrutiny, contribution evaluation, literature positioning, and recommendations across domains whenever a user requests reviews, analyses, or summaries of research publications.