Agent Skills Pack
https://github.com/yizhiyanhua-ai/agent-skills- Category
- AI Agents
- Rank
- No. 2069Tools index
Previous survey · No. 2064 ·
- Pricing
- Open Source
- Type
- TOOL
- Builder
- yizhiyanhua-ai
- GitHub
- 6 stars
- Date
About
Curated collection of Claude Agent Skills covering meeting notes, study notes, email replies, family budgets, and photo organization.
What it does
Agent Skills Pack is a reference library for teaching Claude repeatable, specialized workflows through self-contained instruction bundles. The supplied implementation also includes a command-line video pipeline that builds storyboards, gathers visual assets, analyzes music, previews scenes, and renders vertical videos with Remotion.
Why it's ranked here
It is strongest as a pattern library with one substantial working example. The video code covers orchestration, asset fallback, audio processing, configuration, and explicit errors. Readiness is uneven: another video system remains a design, some validation and preview work is unfinished, and the repository itself warns against relying on these demonstrations without testing.
What's good
The skill format stays simple: metadata plus plain-language instructions, with optional scripts and resources kept together. Users can install grouped skills through Claude Code, use examples in paid Claude.ai plans, or upload custom skills through the API. The implemented video workflow exposes separate generation, asset, preview, render, status, and full-pipeline commands, which supports inspection and partial reruns.
Tradeoffs
The collection mixes open-source material with source-available document skills, so reuse rights vary by component. The video workflow needs external skills, Node.js, Remotion, FFmpeg, Python packages, and potentially several API keys. Asset resolution validation is marked unfinished, single-scene preview rendering remains a placeholder, and failed asset retrieval can leave incomplete scene mappings.
How to use it well
Use it when learning the skill format, prototyping Claude workflows, or adapting the implemented video pipeline with time for environment setup and verification. Install only the relevant skill set, inspect intermediate storyboards and assets, then test rendered output before depending on it. It does not replace production validation, dependency management, licensing review, or a finished technical-explainer video system.
Technical notes+
README.md defines each skill around a SKILL.md with required name and description frontmatter, and documents Claude Code plugin, Claude.ai, and Claude API surfaces. mv-skill/scripts/cli.py uses argparse subcommands for generation, asset fetching, preview, rendering, environment status, full runs, and medleys. mv-skill/scripts/config.py centralizes output dimensions, timeouts, styles, API keys, and external skill locations, but creates cache, output, and session directories during import. mv-skill/scripts/asset_manager.py shells out to companion skills and contains a TODO for real resolution checks. mv-skill/scripts/renderer.py copies assets into Remotion public storage, writes render props, invokes npx remotion render, and leaves single-scene still rendering as a TODO. mv-skill/scripts/music_analyzer.py, mv-skill/scripts/music_merger.py, and mv-skill/scripts/dj_effects.py wrap FFmpeg or FFprobe, with optional librosa analysis. docs/plans/2026-01-23-tech-video-skill-design.md explicitly describes the Manim system as designed but awaiting implementation.
Observed
- License
- Many skills use Apache 2.0; document creation and editing skills are source-available rather than open source.
- Implementation language
- The supplied MV command-line implementation is Python, with rendering delegated to a Remotion project.
- Install surface
- Claude Code supports marketplace registration and separate document-skills or example-skills plugin installation.
- Interfaces
- Supports Claude Code plugins, paid Claude.ai plans, the Claude API, and a Python command-line interface.
- Platform requirements
- The MV workflow checks for Node.js and FFmpeg and requires Python packages; its setup notes include a macOS FFmpeg command.
- Structure
- Skills are self-contained folders containing instruction metadata, with optional scripts and resources.
Read from README.md, docs/plans/2026-01-23-mv-skill-design.md, docs/plans/2026-01-23-tech-video-skill-design.md, mv-skill/scripts/cli.py, mv-skill/scripts/config.py, mv-skill/scripts/__init__.py, mv-skill/scripts/renderer.py, mv-skill/scripts/dj_effects.py, mv-skill/scripts/exceptions.py, mv-skill/scripts/music_merger.py, mv-skill/scripts/asset_manager.py, mv-skill/scripts/music_analyzer.py.
What it can do
Generate meeting notes from recorded discussions
Meeting recording or transcript → Structured meeting notes with key points and action items
Create study notes from educational content
Textbooks, lectures, or study materials → Organized study notes and summaries
Compose email replies
Incoming email and response context → Professional email responses
Create family budget plans
Income, expenses, and financial goals → Detailed family budget with categories and allocations
Organize photo collections
Digital photos and images → Categorized and tagged photo library
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.