Skill UI

This skill provides advanced guidelines for designing robust and effective tools specifically for autonomous AI agents. It shifts the focus from traditional developer APIs to contracts optimized for Language Models (LLMs). Key topics covered include treating tool descriptions as prompt engineering, implementing namespacing, applying the Consolidation Principle, and utilizing Architectural Reduction patterns (like file system access) to maximize agent reliability and workflow complexity.

An advanced reliability wrapper for AI agents that monitors, manages, and recovers from tool call failures. It automatically handles common issues such as truncated JSON responses, API timeouts, rate limits, and mid-chain process breakdowns. By implementing sophisticated failure classification, checkpointing, and learning mechanisms, Guardian ensures complex, multi-step agent workflows remain stable and resilient.

Provides opinionated, evolving constraints to guide agents when building interfaces. Helps identify high-impact design problems and propose implementation-ready improvements.

This skill acts as a routing layer for UI development tasks. It helps agents select the smallest useful UI skill context via the ui-skills CLI before implementation. It ensures efficient context loading by identifying the task category and loading only necessary skills, avoiding unnecessary overhead.

This skill humanizes natural-language memory files like CLAUDE.md, AGENTS.md, and docs by removing AI-isms and adding burstiness. It offers deterministic and LLM modes, preserves code blocks, URLs, paths, and headings, and creates backups. Ideal for cleaning up AI-generated content in configuration files and documentation.

This skill enables agents to search and fetch high-quality, free professional photography from Unsplash for websites, apps, and marketing materials. It helps replace generic placeholders with premium visual assets, ensuring a modern aesthetic through intentional keyword search, filtering, and dynamic image resizing via API.

Fetch real U.S. residential property data via the Zillapi REST API: resolve addresses, Zillow URLs or zpids, then pull valuations, rent estimates, price and tax history, schools, photos and agents, or search for-sale, for-rent and sold listings.

This skill captures and persists critical user decisions, project constraints, and architectural choices across multiple sessions and agents. It saves this context into a structured memory database (`mdbase`), ensuring that future development steps or agents can retrieve the original intent without re-deriving it from scratch. Ideal for documenting project rules, tech stacks, and design rationale.

LWC is a durable memory system for coding agents that preserves project decisions, code structure, and verified context across sessions. It utilizes a physical Wiki document graph and a CodeGraph index to enable source-grounded recall. This tool helps agents search, update, and maintain knowledge without losing critical information during future coding-agent interactions.

Voice agents enable natural human-AI conversation through speech recognition and synthesis. This skill covers low-latency architectures like speech-to-speech pipelines and STT-LLM-TTS workflows. It includes optimization strategies for sub-800ms response times, voice activity detection, and integration with tools like Deepgram, ElevenLabs, and OpenAI. Ideal for building production-ready voice assistants with natural interaction flows.

Expert in building voice AI applications - from real-time voice agents to voice-enabled apps. Covers OpenAI Realtime API, Vapi, Deepgram, ElevenLabs, LiveKit, and WebRTC. Focuses on low-latency, production-ready voice experiences.

Build real-time conversational AI voice engines using async worker pipelines, streaming transcription, LLM agents, and TTS synthesis with interrupt handling and multi-provider support. Ideal for creating voice assistants, customer service agents, and voice-enabled applications that process streaming audio and integrate multiple AI providers effectively.