Skill UI

Implements a comprehensive defense-in-depth framework for autonomous AI agents that interact with external tools. It mitigates risks like prompt injection or tool misuse by enforcing strict tool allowlisting, binding tools to least-privilege identities, requiring human-in-the-loop approval for high-impact actions, and utilizing policy enforcement (like NeMo Guardrails) to bound the blast radius and ensure compliance.

This skill structures a highly controlled, bounded development life cycle for AI agents. It mandates a structured three-phase workflow: Planning (Spec), Implementation (Build), and Verification (Review). It is essential for building complex features from scratch, especially when dealing with ambiguous requirements, high risk, or needing explicit human approval gates to ensure the resulting code meets all defined specifications.

This skill provides a systematic, four-phase workflow for diagnosing and recovering from AI agent failures. It guides the agent to first capture the failure state, diagnose the root cause (e.g., loop, context overflow, state drift), apply minimal containment recovery actions, and finally generate a detailed introspection report, minimizing the need for human intervention.

This comprehensive guide outlines a rigorous, human-in-the-loop engineering workflow for leveraging advanced AI agents to complete complex implementation tasks. It mandates defining clear completion criteria, decomposing work into manageable units, and strategically routing models (Haiku, Sonnet, Opus) based on task complexity. The workflow emphasizes continuous evaluation, regression testing, security review of invariants and edge cases, and strict cost discipline to ensure robust and reliable software development using AI assistance.

This skill guides the architectural design of an Itô Data Atlas agent for advanced quantitative research. It facilitates market discovery, constructs candidate prediction-market baskets, drafts necessary parameter changes, and integrates human-in-the-loop review. It is strictly for workflow and architectural planning, and does not execute live trading orders.

Build and evolve React/Next.js agentic web apps that pair CopilotKit with Microsoft Agent Framework agents hosted in Microsoft Foundry over the AG-UI protocol. Covers wiring choices, native Foundry primitives, tools, human-in-the-loop approvals, generative UI and shared state, local run, azd deployment, and event-stream debugging.

This comprehensive guide helps authors navigate the complex decision of where to submit their Human-Robot Interaction (HRI) research. It outlines critical criteria, including the robot-embodiment and human-in-the-loop tests, to determine the correct venue (e.g., HRI, CHI, ICRA/IROS). Furthermore, it details the five specialized tracks within HRI and advises authors on matching their primary technical contribution to the most suitable academic domain.

This orchestrator manages the entire end-to-end pipeline for preparing technical talks, transforming raw materials—whether code commits (REX mode) or conceptual articles—into polished revision sheets. It sequentially runs through multiple stages (Research, Concepts, Scripting, Revision) with human-in-the-loop checkpoints, ensuring a structured and comprehensive content creation workflow.

This skill manages the human-in-the-loop verification phase of an agent workflow. It ensures that critical artifacts—such as plans, specifications, or migrations—receive mandatory, structured human sign-off before being considered complete or shipped. It collects severity-graded, machine-parseable feedback (e.g., BLOCKER, MAJOR) and acts as a strict gate, preventing completion until all objections are resolved and the review is formally signed off.

Design patterns for building autonomous coding agents, covering tool and function calling APIs, permission and approval systems, browser automation, and human-in-the-loop workflows inspired by Cline and OpenAI Codex.

Expert guidance for building production-grade stateful, multi-actor AI agents with LangGraph, covering graph construction, state management, cycles and branching, checkpoint persistence, human-in-the-loop patterns, and the ReAct agent workflow.

A human-in-the-loop studio for iteratively tuning AI outputs or reviewing complex media/prose. It allows users to directly manipulate visual parameters (color, typography, layout) or annotate documents in real-time, moving beyond static comparison grids or vague prompts. Perfect for designers and content creators who need hands-on, visual feedback.