Agent Skills Hub

always-works-testing

Mandatory execution-based validation for all software implementation tasks. Ensures code works through empirical verification before confirmation.

wrap

Automated session cleanup and documentation tool. Proactively updates CLAUDE.md, detects automation patterns, extracts insights, and organizes pending tasks.

text-to-speech

Expert Kokoro TTS implementation skill for real-time, secure, and offline voice synthesis in JARVIS-style assistants. Features streaming output, prosody control, and performance-optimized audio generation.

claude

Anthropic Claude AI models for high-performance coding, large-context analysis, and GUI interaction.

code-reviewer

Conduct automated code reviews for local changes or remote GitHub Pull Requests. It analyzes code for correctness, maintainability, and standards using git and gh CLI integration.

claude-code

Standardized skill for Claude Code agents to dynamically query OpenRouter model recommendations and metadata via the Claudish CLI.

Extract structured data from unstructured files (PDF, PPTX, DOCX...)

Implement LlamaExtract for robust structured data extraction from PDF, DOCX, and PPTX files using Pydantic schemas.

flow-nexus-neural

Train and manage neural networks in distributed E2B sandboxes using the Flow Nexus platform, supporting custom architectures like Transformers, LSTMs, and GANs.

low-complexity

Enforces low Cognitive and Cyclomatic complexity in all code. Automatically maintains readability, modularity, and maintainability by preventing complex functions during development.

memory

Manage project SSOT, memory, and cross-tool search. Guardian of decisions.md and patterns.md for Claude Code. Use for context retention, memory synchronization, and decision tracking.

hf-mcp

Connect your AI agent to the Hugging Face Hub via MCP. Search models, datasets, and papers, manage repos, run cloud compute jobs, and invoke Gradio Spaces as functional AI tools.

Create New Skills

A framework for creating reusable Claude Code agent skills, following best practices for directory structure, progressive disclosure, and multi-file patterns.