What problem does it solve? Building agentic AI features into applications normally requires implementing your own orchestration, tool invocation, and session management. This Skill provides the GitHub Copilot SDK, which exposes the production-tested agent runtime behind Copilot CLI so you can invoke planning, tool calls, and file edits programmatically. ## Core Features & Use Cases - Programmable Agent Sessions: Create sessions, send prompts, stream responses, and persist conversations across restarts in Python, TypeScript, Go, or .NET. - Custom Tools and Agents: Define tools with schemas and handlers that Copilot invokes during reasoning, plus custom agent personas and system messages. - MCP Server Integration: Connect sessions to Model Context Protocol servers such as the GitHub MCP server for repository, issue, and PR access. - Use Case: Build an interactive CLI weather assistant where Copilot decides when to call your get_weather tool, streams the answer token by token, and maintains multi-turn context. ## Quick Start Ask the agent to create a Copilot SDK session in TypeScript that sends a prompt with streaming enabled and prints the response.