Skill UI

This skill extracts structured data, such as text, tables, or attribute values, from rendered web pages. It supports both running pre-saved templates and conducting one-shot DOM queries. Crucially, it integrates mandatory AIDefence gates for PII scanning and prompt-injection detection, ensuring data safety and reliability before the data reaches the model.

This tool automates the process of filling out web forms by accepting a structured field map (field name to value). It intelligently resolves field selectors either by leveraging stored browser templates for known sites or by analyzing the current page's accessibility snapshot. It supports submission and manages PII redaction.

This utility formats prompts for multiple Large Language Model providers (including Anthropic, OpenAI, Gemini, and Ollama) using standardized chat templates. Furthermore, it enables the creation and utilization of an HNSW index for advanced context retrieval, making it ideal for building robust Retrieval-Augmented Generation (RAG) pipelines.

This comprehensive guide helps elevate a UI from merely functional to polished, purposeful, and aesthetically coherent with its domain. It dictates pre-coding design choices, such as defining the audience, tone (e.g., minimal, industrial), and purpose. It provides rigorous guidelines on establishing visual hierarchy, typography, color palettes, and interactions, ensuring the final product feels intentional and professional, rather than generic or templated.

A powerful tool designed to scaffold sophisticated, custom AI agent harnesses using the metaharness CLI. It supports various templates (e.g., coding, devops, marketing) and AI hosts (e.g., Claude, Codex). Crucially, it operates with maximum safety, defaulting to a dry-run mode and refusing to write to the calling repository root, ensuring controlled and predictable development workflows.

Calculates a weighted similarity score between two complex 'harness fingerprints' (containing genome data and scores). The score is a composite of cosine similarity (numerical), categorical agreement, and Jaccard index (topological). This tool is essential for architectural decision-making, such as determining whether to scaffold a new system or fork an existing one, ranking candidate templates, or detecting architectural drift across different teams.

This tool automates the creation of new, sequentially numbered database migration files, including both 'up' (forward) and 'down' (rollback) SQL scripts. It intelligently selects the appropriate SQL template based on the migration name (e.g., create_, add_, drop_) to ensure correct schema evolution and includes steps for metadata storage and pattern searching.

This skill runs a single, simulated tick of a defined business pod (Sales). It is designed for dry-run testing, allowing users to validate inputs, resolve agents, and reserve budgeted resources without invoking real LLM workers. It handles template loading, schema validation, budget management, and emits structured status reports for subsequent workflow consumption.

A comprehensive tool for authoring and managing complex workflows. It supports two primary methods: persistent MCP templates for long-lived, human-gated, resumable pipelines, and native JavaScript scripts for advanced, code-driven orchestration, allowing for complex fan-out, parallel processing, and structured data aggregation.