Shape agent capabilities into your workflow.

Use this Claude Code skill when a human-facing result benefits from sections, comparisons, or visual structure. It creates one self-contained HTML file with inline CSS, JavaScript, and SVG, using a warm editorial style and responsive layout; the file works offline and is opened in a browser. Short answers, linear explanations, and code-only output remain Markdown.

This Claude Code skill collects references in three sequential layers—UX patterns, visual direction, and mood—and fills a Figma Design file with screenshots, sources, and links as it runs. Claude in Chrome MCP is required; Figma MCP is optional for automatic board creation. It suits designers who want live examples to shape a new project's direction, but it is a reference-discovery workflow rather than a design generator.

Use this Japanese-first Claude Code skill to design, implement, or review presentation dashboards with HTML and ECharts. It organizes requirements through 5W1H, uses a six-part grid, recommends four basic chart types, and checks contrast and accessibility with a 26-item review flow. The scope is presentation dashboards rather than exploratory analysis, and the included guidance and templates are primarily Japanese.

This standalone agent skill scans local files, lockfiles, processes, persistence points, containers, and CI surfaces for indicators from documented npm and PyPI worm campaigns. Its key safety rule is to stop persistence and quarantine evidence before revoking credentials, then release a dependency-ordered rotation plan only after a clean second scan. It supports macOS, Linux, Windows, WSL, containers, and CI, but remains an incident-response runbook rather than a guarantee of safety.

Mentor adds an active teaching mode to Claude Code: it asks questions, offers progressive hints, explains concepts, and can patrol project changes while leaving implementation to the learner. It supports Socratic, tutor, or mixed teaching, multiple languages and humor styles, plus persistent user and project settings. Use it for deliberate practice and accountability, not for having the assistant write functional code.

Use this Claude Code skill to run Screaming Frog headless crawls, query existing Derby databases, diff crawl results, and render client-facing HTML audit reports. The bundled checks cover broken links, redirect chains, titles, H1s, thin content, page size, noindex pages with inlinks, and orphan candidates. Fresh headless crawls require a paid Screaming Frog license; queries and reports can use existing GUI-produced crawls.

Resume Writing Skill turns an unstructured career dump or an existing resume into three artifacts: a clean Markdown resume, a moderncv-banking LaTeX version, and a personalized teaching handbook. It diagnoses whether the problem is packaging or thin experience, compares the target role with current US job descriptions, and flags numbers that need verification instead of silently carrying them forward. It is scoped to US-market resumes, not cover letters or interview preparation.

Agentic AI Security applies a five-category framework while an LLM-based agent is being designed or coded. It checks privilege and identity boundaries, configuration and segmentation, behavior risks such as specification gaming, structural risks such as tool injection and cascading failure, and accountability gaps in monitoring. The coverage spans design through operation, but excludes infrastructure hardening, model-training security, and regulatory compliance.

Interview Sim turns an interview protocol and a real or fictional persona into a text transcript with calibrated pacing, hesitations, self-corrections, and natural follow-up probes. It begins with an intake to align duration, language, scenario, interviewer style, and fidelity, then runs a self-check for persona consistency and fabricated claims. The output is text only, with one interviewer and one interviewee; it is not an audio transcription tool.

Iterate is a meta-skill for improving other Claude Code skills through run evidence. An opted-in target records structured friction events, and Iterate accumulates them across runs before asking a read-only agent to propose an SKILL.md diff in a pull request. It never auto-merges, weakens deny rules, skips safety hooks, or changes files outside the target skill. Friction state is isolated per project.

Win-C-Cleaner guides Windows 10 and 11 C-drive cleanup through eight stages, from temporary files and hibernation data to component storage, duplicate drivers, OEM software, and recovery packages. It requires an administrator PowerShell, validates deletion paths, defaults the first stage to dry-run, and adds explicit confirmations or backups to riskier operations. Some component-store cleanup cannot be rolled back, so the staged workflow matters.

OSS Contributor helps aspiring AI engineers find and track open-source work instead of browsing random good-first-issue labels. It searches configured repositories through the GitHub REST API, filters candidates with a five-point verification, and monitors claim comments, pull-request status, and escalating deadlines. A small setup script stores per-project state in `.claude/oss-config.json`; the runtime uses only Python’s standard library.

Reality Citation Audit checks a LaTeX paper’s bibliography end to end. It verifies DOI metadata through Crossref and related sources, resolves anonymous authors and published versions, finds missing or dead entries, and tests whether citations actually support the surrounding claims. It can merge duplicate keys, rewrite references, rebuild the paper, and report remaining issues. A `.tex` project, `.bib` file, network access, Python, and LaTeX tools are required.

Google Map Voices collects public Google Maps reviews with SeleniumBase, supports several sort modes, merges duplicates, and saves fields such as text, stars, names, and dates to JSON. You can pass that file to Claude or GPT using the included analysis guide to extract positive quotes, pain points, and a concise insight. It is intended for small reconnaissance jobs, not paid bulk scraping or automated review replies, and Google’s UI changes may require maintenance.

DailyForge turns the last 24 hours of Git activity across multiple Unity repositories into a narrative report for Discord. It enriches commits with Unity-specific context and supports whole-project, personal-commit, or anonymous team scopes without commit counts or rankings. Configure the provider, repositories, recipients, and webhook, and notify affected developers in writing before processing their data.

Use avis to place comments directly on a web page and let Claude see them in their original context. The skill also supports asking the agent to annotate the page for you, making it useful for design or content feedback tied to specific locations. It requires a browser MCP server; perch is the documented macOS reference backend.

Add a floating toolbar to a React 18+ prototype so designers, PMs, and stakeholders can switch component alternatives in place. The toolbar highlights the component under each option, keeps selections and layout state in local storage, and shows only versions whose wrapped component is in the viewport. It has no new npm dependencies; the page-side toolbar is local-only by default, while production use relies on the companion extension or an explicit override.

This repository is a Claude Code marketplace for Glitchwerks projects. Its current listing is claude-wayfinder, which helps select agents and skills through typed, auditable scoring; a prospector plugin is listed as upcoming rather than available. Marketplace entries pin source commits so installs are reproducible, making the repo useful for users who want a small, version-pinned plugin catalog.