lepinkainen

Add beui-style loading animations (spinner, bouncing dots, bars, dot-matrix, dither, terminal/ascii spinners, shape morph, comet, scramble text, metaballs, newton's cradle, helix, percent counter) to any project as vanilla HTML/CSS/JS — no React, no build step, no dependency. Use this skill whenever the user wants a loader, spinner, loading indicator, progress animation, or "loading..." state in a non-React project — server-rendered templates (Go, Rails, Django, Flask/Jinja, PHP), HTMX, Web Components, static HTML, or plain JS apps — even if they don't mention beui by name. Also use it when the user asks to port or copy a beui motion component.

Audit a codebase for health issues — oversized files, duplicated logic, dead code, unused parameters, test coverage gaps, and documentation/config drift — and report findings ordered by severity. Use this skill whenever the user invokes "/check-code-health", asks about "code health", "tech debt", "code quality audit", "find dead code", "what needs cleanup", or wants a maintenance overview of the repo.

Update project dependencies safely — list outdated packages, upgrade minors/patches in bulk and majors individually with changelog review, verify with build and tests, and commit in clean chore(deps) commits. Use this skill whenever the user says "update dependencies", "upgrade deps", "bump packages", "are my dependencies outdated", "update to the latest version of X", or mentions dependabot/renovate-style maintenance. Not for adding a brand-new dependency.

Diagnose and fix a bug or GitHub issue with test-backed verification — reproduce, root-cause, fix, and guard with a red-green or regression test so it can't silently return. Use this skill whenever the user says "fix this bug", "there's a bug in", "this is broken", "fix issue

Get a failing build back to green — run the project's build, test, and lint commands, then fix every reported compile error, test failure, and lint violation until all pass. Use this skill whenever the user invokes "/fix-build", says "the build is broken", "fix the build", "it doesn't compile", "CI is red", "make the tests pass", or pastes build/compiler errors asking for them to be resolved.

Generate or update CLAUDE.md for a codebase the thorough way — discover architecture, workflows, and project-specific conventions, merge with existing AI-agent instruction files, and iterate with the user. Use this skill whenever the user invokes "/initialise", says "set up CLAUDE.md", "update the CLAUDE.md", "initialise this repo for Claude", or wants AI-agent onboarding docs written for a project. Prefer this over the built-in /init when the user asks for the fancy/thorough version.

Analyze a GitHub issue and write an implementation plan into the issue itself — without implementing anything. Use this skill whenever the user invokes "/plan-github-issue", says "plan issue

Cut a release from Conventional Commits history — generate or update the changelog, decide the semver bump, update version files, and create an annotated git tag. Use this skill whenever the user says "cut a release", "make a release", "prepare a release", "bump the version", "tag a version", "update the changelog", or asks "what should the next version be". Also use it when the user names a specific version to release ("release v1.4.0").

Analyze Go test files (*_test.go) for quality issues, anti-patterns, and code smells. This is a Go-specific tool that checks for external dependencies, complexity, flaky patterns, and testing anti-patterns. Use when users ask to check Go test quality, analyze Go tests for issues, find test anti-patterns in Go code, find flaky Go tests, or improve Go test reliability.

Analyze codebase health - large files, test coverage gaps, duplicate code, dead/legacy code, and documentation issues. Use when asked to scan, audit, or assess code quality, find refactoring targets, or identify technical debt.

This skill should be used when the user asks questions about their Plex media library, such as "what movies can I watch on Plex", "find romantic movies under 100 minutes", "show me Tom Cruise films from the 90s", "how many unwatched episodes of Always Sunny do I have", "give me a British comedy show I haven't watched", or "shows with episodes under 30 minutes that I've started". Use whenever user mentions Plex in combination with movies, TV shows, or media queries.

Verify TUI client changes end-to-end by driving build/lurker-tui in an isolated tmux session against a seeded backend. Use when verifying changes under cmd/tui/ (rendering, keybindings, sidebar, buffer switching, mouse support) at the real terminal surface. Not for backend-only changes and not for running unit tests (task test-tui is CI's job).

Verify web UI changes end-to-end by driving the built frontend against a seeded backend with Playwright. Use when verifying changes under web/src/ (rendering, keyboard shortcuts, sidebar, markers, WebSocket-driven behavior) at the real browser surface. Not for backend-only changes (use curl against the API instead) and not for running unit tests.

Add a new feed provider to feed-forge. Use when the user asks to "add a provider", "support a new feed source", "wire up <site> feed", "create a new feed source", or anything else that means a new entry in internal/<name>/ that implements providers.FeedProvider and self-registers with the global registry.