Skill UI

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.

This guide provides a systematic, safe workflow for refactoring rapidly developed or AI-generated fullstack applications. It focuses on identifying and removing dead code, fixing broken imports, and consolidating repetitive logic while strictly maintaining existing API contracts, routes, and core functionality. Use this before launch to ensure a stable, maintainable production baseline, following a 'surgery, not demolition' philosophy.

This skill helps identify explanatory attributes or hypotheses for variations in process metrics such as duration, waiting time, utilization, and rework. It involves checking metadata correlations, statistical evidence, and cohort comparisons to distinguish competing explanations. The method emphasizes evidence-based analysis, avoiding silent assumptions, and framing findings as testable hypotheses rather than definitive causal claims. It supports structured investigation of process efficiency and bottlenecks using specific analytical tools and frozen baseline scopes.

Analyze prompt structure to maximize LLM cache reuse, then measure input cost and latency. Covers stable prefixes, user/workspace boundaries, provider-specific cache counters and a cost comparison calculator for OpenAI, Claude and Gemini.

Manage stable domain terminology and separate business semantics from architecture decisions. Create CONTEXT.md for ambiguous terms or repeated explanations, and route irreversible decisions to existing Architecture Decision Records. Prevents implementation drift by clarifying language and ensuring traceability to code or business evidence.

Inventory and govern a project's test assets. Connect requirements, business rules, risks, bugs, and cross-boundary contracts to stable TEST-IDs and executable evidence in the repository's existing test registry or TESTS.md. Use for test inventory, test gap analysis, unit/integration/contract/E2E/smoke classification, bug-to-test regression protection, consumer/provider contract testing, test-necessity review, and pre-delivery evidence review.

A public-safe preliminary screen for challenges to the impartiality of an AI evaluator, LLM-as-a-judge, or decision-support agent based on prior outputs, suspected bias, unstable reasoning, or self-reinforcing memory. Classifies evidence and recommends proportionate remedies.

This skill ensures browser automation reliability by anchoring actions to stable DOM selectors and semantic roles rather than screen coordinates. It helps workflows survive layout shifts, A/B tests, and viewport changes. Users should inspect the DOM, prioritize data-testid attributes, verify selectors match exactly one element, and wait on state instead of fixed time. It includes error handling for modals and navigation races.

Shared vocabulary and principles for designing deep modules with small interfaces, clean seams, and testable, AI-navigable codebases.