Skill UI

This pattern addresses the fundamental limitation of context window size by treating the filesystem as a persistent, external memory. Agents can write large outputs, plans, or findings to files, offloading them from the active context. Instead of carrying all information, agents use targeted search and retrieval tools to load only the necessary context on demand, drastically reducing token bloat and improving performance over long, complex tasks.

A deep expert system for Anthropic's Claude Code CLI. This skill maximizes productivity by managing advanced features like hooks, persistent context via CLAUDE.md, memory management, sub-agent orchestration, and integrating with local development ecosystems. Ideal for complex coding tasks, debugging, and advanced workflow automation.

Evaluates incoming requests, decides if specialized skills are needed, orchestrates optimal combinations, and logs successful pairings in agent-memory while guarding against overusing complex tools on simple tasks.

Manages and coordinates complex workflows involving multiple AI agents and tasks. It efficiently distributes work across multiple workers, handles runtime errors, and intelligently merges disparate results into a unified output, making it ideal for large-scale, distributed AI computations.

This autonomous skill factory converts raw, ambiguous human input—such as messy transcripts, documents, screenshots, or vague workflows—into complete, production-ready, cross-platform agent skills. It acts as a powerful intermediary, deriving complex intent and automating the creation of reusable software, bridging the gap between human articulation and technical specifications. Ideal for automating repetitive tasks and building custom agent capabilities with minimal coding knowledge.

Run Hugging Face models directly in JavaScript/TypeScript for text, vision, and speech tasks without backend servers, using Transformers.js in Node.js or modern browsers with WebGPU, quantization, and pipeline abstractions for efficient inference.

This harness transforms standard code environments into fully persistent, autonomous agent systems. It leverages native features like scheduled tasks (Crons), remote dispatching, and structured memory storage (MCP) to enable continuous, self-directing automation. Ideal for building AI assistants that require long-term context, recurring workflows, and complex, multi-step operations, eliminating the need for external agent frameworks like AutoGPT.

A cross-protocol platform for discovering and searching 18,000+ MCP servers and AI agents across six or more registries. It allows users to find the necessary endpoint for specific tasks (like database access or file conversion) by searching across MCP, A2A, and agents.txt protocols simultaneously.

A robust, production-tested pattern for coordinating multiple specialized AI agents. The orchestrator decomposes complex tasks, routes them to the most suitable specialist, prevents redundant or conflicting work (anti-duplication), and enforces mandatory quality gates (e.g., running tests, checking file diffs) to verify results before marking a task as complete. It also includes heartbeat monitoring for task progress.

This skill provides comprehensive interaction with the DeerFlow AI agent platform via its robust HTTP API. It is designed for managing complex, multi-step AI workflows, including deep academic research, code execution, web browsing, and file analysis. Users can initiate and continue conversation threads, manage agent memory, upload documents (PDF, etc.), and delegate complex research tasks to the autonomous agent system.

Switches the AI persona into a Chinese-style mama nagging narrator that sticks to core behavior guardrails while delivering increasingly intense motivational scripts; useful when you need a relentless reminder that you must research, verify, and finish tasks before asking for help.

This skill guides users on setting up and running LLM-as-judge evaluations using Arize. It covers defining evaluators (including prompts, classification choices, and models) and executing tasks against real data spans or experiments. It supports key metrics like hallucination, faithfulness, and correctness, and facilitates continuous monitoring for comprehensive LLM performance analysis.