Skill UI

A mechanism for spawning hierarchical, nested sub-agents up to five levels deep, designed specifically for deep task delegation while managing context limitations. Instead of parallel execution, this ensures that each layer of decomposition receives a fresh context window, preventing the parent agent's context budget from being overwhelmed by complex, multi-step workflows. Ideal for research, verification, and synthesis processes.

Pilot Protocol is an open-source overlay network providing AI agents with first-class network citizenship. It offers permanent virtual addresses, encrypted P2P communication, NAT traversal, and a structured app store. Use this skill when agents require stable networking across cloud changes, need direct, secure messaging without shared accounts, or must integrate structured external data services (e.g., live crypto prices) via local, typed capabilities.

This skill provides guidance on implementing common architectural patterns and best practices within the PyTorch framework. It is designed for developers building sophisticated deep learning models, helping to improve structure, efficiency, and code maintainability when working with neural networks.

A structured framework for managing complex, iterative, and high-stakes decision processes (e.g., stochastic optimization, ensemble comparison, model rollouts). It mandates the creation of a detailed, auditable 'ledger' that records every prior result, new input, search space metric, and decision mark. This tool enforces rigorous evidence trails and defaults to safe, dry-run modes, ensuring that confidence is built through verifiable, step-by-step proof rather than assumed certainty.

This tool manages agent memory stored in the RVF (Ruflo Vector Format). It facilitates the backup, transfer, and sharing of complex agent knowledge, including vector embeddings, metadata, and causal relationships, across different projects and platforms. Users can list, export, import, migrate, and analyze memory statistics to ensure knowledge continuity.

This tool assesses whether coding agents adhere to defined skills, rules, or agent definitions. It automatically generates scenarios with varying prompt strictness, executes the agent, and generates a comprehensive report detailing the compliance rate of tool calls against predefined workflow specifications using advanced LLM classification.

This skill provides regulator-grade feature attribution for complex trading signals (LSTM/Transformer). By constructing a feature-contribution graph and running single-entry PageRank, it identifies the top-K features that most drove a prediction. This is crucial for compliance with regulations like the EU AI Act and SEC Reg-AI, providing a robust, auditable audit trail for algorithmic trading systems.

This skill generates high-dimensional vector embeddings (384-dim) from text, code, or documents using the ruvector package. It utilizes the ONNX all-MiniLM-L6-v2 model, making it ideal for implementing advanced semantic search, similarity comparison, and data clustering. The resulting vectors are normalized and can be efficiently stored and indexed using HNSW, significantly enhancing retrieval accuracy in RAG and knowledge systems.

A comprehensive vector search plugin offering two distinct paths: large-scale corpus search using HNSW (supporting RaBitQ quantization for memory efficiency) and a fast WASM-backed router for high-priority hot patterns. Ideal for implementing Retrieval-Augmented Generation (RAG), semantic search across massive datasets, or routing queries to specific, limited patterns.

A comprehensive, multi-stage workflow for enhancing and structuring raw footage using advanced AI tools. It covers the entire pipeline—from initial recording (Screen Studio) through AI planning (Claude/Codex), deterministic cutting (FFmpeg), compositing (Remotion), asset generation (ElevenLabs/fal.ai), and final human polish (Descript/CapCut). Designed for users creating vlogs, tutorials, or polished corporate videos.

Manage the lifecycle of WebAssembly (WASM) agents, providing a secure and isolated environment for running untrusted code and conducting experiments. This tool allows users to create agents, interact via prompts, equip them with specific tools, manage sandboxed files, and export the configuration, all while guaranteeing full isolation from the host system.