Skill UI

A high-performance utility for executing simple, deterministic code transformations (e.g., removing console logs, converting var to const). This engine leverages a WASM core to perform structural refactoring sub-millisecond, without calling any expensive LLM APIs. Ideal for CI/CD pipelines, automated bulk refactoring, and cost-sensitive development workflows where determinism and speed are paramount.

This tool routes tasks through a multi-tier system, specifically partitioning tasks into Agent Booster eligible (Tier 1) and standard LLM calls (Tier 2/3). It calculates and reports the estimated cost, leveraging the structural $0 cost of Booster bypasses for simple transforms, helping users optimize spend before running batches of similar tasks.

Automatically analyzes complex AI operations, including token usage patterns, model tier suitability, cache hit rates, and agent redundancy. It identifies potential waste and provides actionable, prioritized recommendations with estimated dollar savings to significantly reduce operational costs and improve efficiency across LLM pipelines.

This skill analyzes historical benchmark run data stored in JSON files. It surfaces key performance drift metrics over time, including win rate, average latency, p99 latency, and LLM-baseline cost. It is crucial for detecting subtle performance degradations or regressions that standard smoke tests might miss, ensuring stability across releases or corpus expansions.

Build a production-ready, AI-powered data collection agent for any public source (job boards, prices, news, etc.). This agent automatically scrapes data, enriches it using free LLMs (like Gemini Flash), and stores the results in databases such as Notion, Sheets, or Supabase. It runs on a free schedule using GitHub Actions and learns from user feedback, making it ideal for continuous data monitoring and tracking.

This guide details integrating FoundationModels for running Large Language Models (LLMs) directly on the device (iOS 26). It covers advanced patterns including text generation, extracting structured data using @Generable, implementing custom tool calling for domain logic, and streaming responses for real-time UI updates. Key features emphasize strong privacy and offline capability, ensuring no data leaves the device.

This skill provides a detailed, side-by-side comparison of advanced LLM agent evaluation frameworks, such as ruflo and HAL. It helps researchers and developers identify capability gaps—including real code execution, full web browsing, and structured file parsing—and prioritize architectural improvements for robust agent benchmarking and performance tracking.

This skill provides a systematic workflow for diagnosing failures in GAIA benchmark runs. Users can analyze task traces, classify failure modes (e.g., Tool Gap, Reasoning Miss, Extraction Bug), identify root causes, and apply targeted fixes to improve AI agent performance. It is essential for optimizing LLM benchmarks and finding performance regressions.

GateGuard is a mandatory pre-tool-use hook designed to significantly enhance the reliability and safety of LLM-generated code and actions. Instead of relying on self-evaluation, it forces the model to perform concrete, verifiable investigations—such as listing all imported files, checking data schemas, or detailing rollback procedures—before any write or destructive command execution. It establishes a three-stage gate (Deny, Force, Allow) to prevent hallucination and ensure context awareness in complex coding environments.

This tool provides deep inspection and auditing capabilities for GEPA (genetic-evolution prompt-adaptation) genomes by utilizing the internal `@metaharness/darwin/gepa` library. It is crucial for advanced AI development pipelines, allowing users to validate the structural integrity of genomes, render the full system prompt derived from a genome, or classify specific failure modes found in an execution transcript. Ideal for CI/CD gating and debugging complex LLM applications.

Provides comprehensive configuration management for local LLM inference using RuVLLM. Users can select models, perform task-specific fine-tuning with MicroLoRA adapters, or implement continuous, real-time adaptation using SONA. This tool guides the entire advanced LLM workflow, covering status checks, configuration generation, and advanced adaptation loops.

This skill provides comprehensive security and monitoring for Large Language Model (LLM) powered trading agents. It helps to audit trade decisions, ensure compliance with risk parameters, and mitigate potential risks associated with automated financial operations.