Helps Claude make deterministic, auditable choices about which agent and skills to use for a given task — replacing prose-scanning agent/skill selection with a typed scoring kernel.
★not rated 2 yesterdayA
tokens not measured
originalMIT
Comprehensive iOS 26 Liquid Glass design system for SwiftUI. Provides API reference, HIG principles, design tokens, motion patterns, accessibility guidance, performance rules, anti-patterns, a pattern library for every common surface, and nine full-screen example apps that compile in Xcode 26.
A plugin containing a Chinese product-planning workflow: writing product requirement documents, syncing them to Feishu, turning them into user stories, and preparing release announcements. Feishu is a collaboration and document platform.
★not rated 2 1mo agoA
tokens not measured
originalMIT
BOUCH Legal OS — 7 Claude skills for UK legal research. Legislation lookup, case law, HMRC/VAT, parliamentary intent, MP research, full legal briefs with OSCOLA citations. MCP-backed via Ledgerhall.
Intelligence layer for Claude skills. Maintains a trust-rated catalog (curated + GitHub topic search), recommends what fits your project, returns ranked lists with pros, cons and customization advice, persists every decision, and migrates skills across 55 supported agent platforms. Ships in three editions: Lite…
★not rated 2 15d agoA
tokens not measured
originalMIT
Claude skills for AI agent and ML reliability: reproduce the eval-to-production gap, catch confidently-wrong outputs, and prove root cause with verified numbers. Start with production-autopsy.
★not rated 2 2mo agoA
tokens not measured
originalMIT
Tutoring skill (teach) for learning a topic interactively — calibration, syllabus planning, and a Socratic/Feynman/drill teach loop. Named after Richard Hamming.
★not rated 2 4mo agoA
tokens not measured
originalMIT
Finds where a Claude Code session wasted tokens (uncached context, context bloat, oversized tool output, redundant reads, wrong-tier routing) and coaches the fix. Adds a live cost/context statusline and a non-blocking end-of-session coach.
★not rated 2 2mo agoA
tokens not measured
originalMIT
Builds a progressive YouTube learning path on any topic. Fans out 8+ search angles with yt-dlp and web search, then reads the transcript and the top comments of every shortlisted video before recommending it. Broad topics are anchored to the field's canonical textbook and delivered with an honest coverage table.…
★not rated 2 1mo agoA
tokens not measured
originalMIT
Hybrid prompt linter — deterministic Node detector for 18 of 20 rules plus an LLM model layer for PR003, PR009, and semantic cases of PR002, PR005, and PR010. Lint, don't rewrite. Markdown or JSON output, suppression comments, baseline mode, conformance corpus (25 cases), CI recipes. Never imposes a framework, never…
★not rated 2 4mo agoA
tokens not measured
originalMIT
Let Claude watch a video -- fully local. Downloads with yt-dlp, extracts auto-scaled frames with ffmpeg, ALWAYS transcribes with faster-whisper and compares against creator-provided captions. Auto-detects your NVIDIA GPU: uses it for video decode (NVDEC) and whisper when present, falls back to a fully working CPU-only.
★not rated 2 26d agoA
tokens not measured
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: