Focused UI polish specialist for an existing product direction. Improves surface depth, optical alignment, radii, icons and interaction details after product and design context are established; use the broader design-engineering workflow for redesign, prototyping or full UX audits.
Prepare and open pull requests for Gentle AI repositories or projects explicitly adopting its issue-first workflow. Use to verify approved issue linkage, branch naming, PR templates, type labels and required checks before submission. Do not impose Gentle AI approval labels or templates on unrelated repositories; use their actual contribution rules.
Split large or review-heavy changes into stacked PRs to main or a feature-branch chain with a tracker. Use when the user requests PR slicing, stacked changes or reviewer-load control, or the effective project delivery strategy requires a split. Define slice boundaries, dependencies, review order and checks without replacing the selected delivery mode or imposing a universal line-count gate.
Trigger: pptx to pdf; extract deck design system; design.md from slides; slidev template from deck; publish deck pages site. Turn a presentation into PDFs, assets, a documented design system and a reusable Slidev template, published with Pages.
Run auditable GitHub roadmap delivery: reconcile issues and dependencies, define independently reviewable units, implement linear commits and PRs, verify CI, record decisions, and synchronize parent roadmaps after authorized merges. Use only when the user explicitly selects enterprise delivery for issue-to-code traceability and audit history; not merely because a codebase is large or uses GitHub.
Investigate consequential claims and decisions through current primary sources, evidence quality, competing explanations, uncertainty, and personalized tradeoffs. Use when the user asks whether a claim is true, whether something is worthwhile, requests a critical comparison or probability, shares promotional content, or wants practical next steps across technology, household, finance, health, and services.
Search GitHub and the open-source ecosystem before substantial implementation, then verify fit, license, maintenance, security, governance and adoption cost and decide USE, CONTRIBUTE, FORK or BUILD. Trigger before implementing non-trivial apps, services, libraries, plugins or feature-level capabilities that may already be solved, including auth, payments, scraping, browser automation, notifications, search/RAG, observability, GitHub automation, media processing/transcription, queues, schedulers, rate limiting and workflow engines. Also use for GitHub/open-source alternatives, repository comparisons, build-vs-reuse decisions, or when deciding whether to adopt, contribute to, fork or rebuild an existing project.
Autonomously assess and maintain GitHub repositories: prioritize roadmap issues and dependencies, implement changes with tests and documentation, review PRs and CI, and handle authorized merges or releases. Use to continue a project, clear a backlog, plan delivery phases, or coordinate authorized agents. Generic fallback only; an explicitly selected Personal or Enterprise delivery mode owns the loop.
Trigger: Google Docs formatting, Google Sheets design, Google Slides layout, Drive document creation and editing. Create polished, native Google Workspace deliverables with structural and visual QA.
Design and evaluate Home Assistant installations, device choices, smart-home protocols, room sensors, automations, voice control, networking, power resilience, backups, and migrations. Use when the user asks about Home Assistant hardware, Zigbee, Thread, Matter, Wi-Fi, Bluetooth, coordinators, blinds, environmental monitoring, local voice, NAS hosting, or phased smart-home purchasing.
Create and triage GitHub issues for Gentle AI repositories or projects explicitly adopting its issue-first workflow. Use to check duplicates, file complete bug reports or feature requests with the correct templates, or guide maintainer review and approval before PR work. Do not impose Gentle AI labels or approval rules on unrelated repositories.
Expose canonical Agent Skills from an MCP server through io.modelcontextprotocol/skills without creating a second behavior source of truth. Use when adding or auditing provider-side skills/list, skills/get, resources/read, manifest integrity, security boundaries, official conformance, evidence levels, and standalone fallback for mixed-client support.
Trigger: bootstrap a public GitHub repository; prepare an open-source project for community contributions; configure GitHub metadata and release workflows for a new public project. Prepare public GitHub repositories for community use.
Trigger: performance audit, analizar rendimiento, mejorar performance, optimizar app, revisar re-renders, bundle size, lazy load. Analiza un repositorio React/Vite/TypeScript, identifica 10 mejoras de rendimiento, crea PR para la más impactante y issues para las demás replicando el formato del repo.
A folder with a SKILL.md file: a name, a description of when to use it, and instructions. Claude loads a skill only when the task matches its description.
How do I use one I find here?
Copy the folder into your project's .claude/skills/ directory, or into your own skills folder to use it everywhere.
What do the warnings mean?
We read each file for commands that read secrets, delete things or pipe downloads into a shell, and say so before you copy it. No warning is not a promise that a file is safe.
Which skills worked for people?
Open a skill to see its discussion. Reports from people and their agents are coming.