Skill UI

The GitHub Copilot SDK empowers developers to embed advanced, agentic AI workflows directly into their proprietary applications. It provides a programmatic interface to Copilot's production-tested agent runtime, allowing you to define complex agent behaviors, manage tool invocation, process file edits, and implement real-time streaming responses. Ideal for building custom, highly intelligent tools and automating complex tasks across multiple languages (Python, TypeScript, Go, .NET).

This advanced agent acts as a highly constrained implementation expert, designed to execute complex, multi-step plans with absolute fidelity. It strictly adheres to the provided workflow, making only specified changes, updating the plan status in real-time, and running tests until completion. Ideal for disciplined software development, detailed project execution, or formal process adherence.

Guides a project planning agent through researching, decomposing, and documenting multi-commit development plans before coding, enforcing autonomous research protocols and template-driven plan outputs for structured PR execution.

This skill guides the parallel assignment of specialized agents when debugging multiple, independent failure domains. Instead of using one agent for all issues, isolate each problem (e.g., a specific test file or subsystem) to its own dedicated agent. This allows agents to work concurrently without sharing context or interfering with each other, significantly accelerating the debugging process and maintaining context purity.

This skill rigorously verifies all claims of completion, whether for a task, a bug fix, or a feature rollout. It prevents false assurances by demanding fresh, multi-faceted evidence, including full test suite results, comprehensive coverage reports, integration status, and validation commands. Use it before declaring a component ready for GO or accepting a sub-agent's success report.

This skill integrates the agent with the Matrix/Element messaging platform. It enables full conversational capabilities, allowing users to receive real-time alerts, search content, and control connected systems (such as cameras) directly from a designated Matrix room. It supports multi-turn interactions, including sending files, reacting, editing, and managing threads.

A powerful pre-action hook designed to enforce rigorous checks before AI agents modify code, create files, or execute destructive commands. Instead of relying on superficial self-evaluation, GateGuard forces the LLM to gather concrete facts (like dependencies, data schemas, and impacted modules), significantly improving code reliability and architectural depth. It provides multiple gates for edits, writes, and destructive bash commands.

This guide provides comprehensive instructions on configuring enterprise-grade access control for Intercom. It covers managing admin roles (Owner, Admin, Agent), implementing OAuth scope-based access for public apps, and enforcing granular permissions at the application layer using TypeScript examples. Essential for building secure, multi-tenant SaaS solutions.

This guide details how to use sophisticated coding agents (such as Codex, Claude Code, or Pi) through a structured bash workflow. It provides methods for complex development tasks, including building new features, refactoring large codebases, and reviewing pull requests (PRs). Key best practices covered include managing pseudo-terminals (PTY), defining focused working directories (`workdir`), and executing tasks in the background for robust, multi-step automation.

This skill helps users design robust and optimal software interfaces (APIs) for a module by applying the 'Design It Twice' philosophy. It generates multiple, radically different architectural designs using parallel agents, allowing for comprehensive comparison of trade-offs (e.g., simplicity vs. flexibility, specialized vs. general-purpose). Use this when you need to explore various structural options or compare module shapes before committing to a final design.

This guide teaches how to build robust, production-grade ReAct agents using LangGraph 1.0's `create_react_agent`. It specifically addresses common agent failure modes, such as preventing infinite loops (recursion caps) and ensuring proper error propagation. Learn to define typed tools using Pydantic, manage state with checkpointers, and set precise recursion limits. Perfect for users migrating from legacy AgentExecutor or building reliable, multi-step tool-calling workflows.

This skill facilitates an interactive Quality Assurance (QA) session. Users can conversationally describe encountered bugs or issues. The agent clarifies the problem, explores the codebase for context and domain language, and then meticulously files durable GitHub issues. It supports both single-issue reports and breaking down complex problems into multiple, independently verifiable tickets, ensuring developers receive actionable, high-quality reports written from the user's perspective.