Skill UI

Diagnoses full-stack failure modes in agent and LLM applications, including wrapper regressions, memory contamination, tool discipline failures, hidden repair loops, and rendering corruption. Produces severity-ranked findings and code-first fixes for production agent systems.

Use this skill to deeply inspect and audit the entire process of local AI coding-agent sessions. It tracks critical operational metrics such as token consumption, cost spikes, tool failures, latency anomalies, and overall health scores. Ideal for debugging unreliable code runs, establishing CI/CD health gates, or comparing different attempts to ensure semantic accuracy.

The Ejentum Reasoning Harness is a comprehensive library of cognitive operations designed to guide complex AI agents through critical thinking tasks. It exposes four specialized harnesses (reasoning, code, anti-deception, memory) that mitigate common failure modes like hallucination and reasoning decay. Instead of a fixed answer, the agent ingests a detailed, executable scaffold—including failure patterns, procedures, and falsification tests—to ensure robust, self-verified output.

A comprehensive, expert-level guideline for building Agentic User Interfaces (UIs). This system emphasizes conversational interactions, clear outcomes, and minimal user controls, allowing users to delegate complex tasks to AI. It provides structured rules covering color tokens, typography, accessibility (WCAG 2.2 AA), and component anatomy for professional developers and designers.

This skill offers a narrative-first interrogation framework for CMOs to validate marketing positioning, ICP, and channel mix before launching campaigns. It answers six critical questions including JTBD, positioning statements, and CAC payback to ensure defensibility. Use it when repositioning or facing rising customer acquisition costs to get a green/yellow/red verdict on strategy viability.

This skill summarizes the current conversation into a structured handoff document, ideal for passing work to a fresh AI agent or starting a new session. It avoids content duplication by referencing existing artifacts (PRDs, plans, issues, commits) using their paths or URLs, ensuring the receiving agent has a concise and actionable overview of the project's state, open decisions, and next steps.

Expert skill for building Model Context Protocol (MCP) servers and tools from scratch. Covers the full development lifecycle—from defining tool schemas and resources to implementation, testing, and publishing to registries. Supports robust patterns using TypeScript and Python for integrating custom capabilities into AI agents.

This skill provides a comprehensive, structured workflow for creating, authoring, and refining new agent skills. It guides users through gathering requirements, drafting structured documentation (SKILL.md), adding utility scripts, and optimizing the skill description for optimal agent recognition. Ideal for system architects and developers building complex AI agent toolsets.

A 12-agent pipeline for writing academic papers across all disciplines. Supports 11 modes including full drafting, outlining, revision coaching, literature review, citation checks, AI disclosure, and conference rebuttal audits, with LaTeX, DOCX, and PDF output.

A domain-agnostic 13-agent pipeline for rigorous academic research on any topic. Eight modes cover full research, quick briefs, paper review, literature review, fact-checking, three-way literature scans, Socratic research dialogue, and systematic review with optional meta-analysis, from question formulation to APA 7.0 reporting.

This is a comprehensive lookup reference for the Mercury MCP server tools. It guides AI agents on how to interact with a virtual team workspace, enabling structured communication for messaging teammates, managing threads and tasks, scheduling complex automations, and performing administrative graph edits. Use this when your agent needs to execute specific team workflows.

This skill guides the creation and maintenance of a `.geminiignore` file. By systematically filtering out machine-generated noise—such as build artifacts, dependency lock files, logs, and caches—it significantly optimizes the AI agent's context window. This process improves processing speed and, critically, reduces token consumption costs (FinOps).