# How to assign and manage Copilot agent tasks from anywhere Source: https://www.youtube.com/watch?v=__RLpdxThEE Recap page: https://rapidrecap.app/video/__RLpdxThEE Generated: 2025-11-20T15:34:54.603+00:00 --- ## Quick Overview GitHub Copilot Agents enable developers to delegate complex, multi-step coding tasks, such as implementing accessibility features or adding fuzzy search, directly through natural language prompts in the VS Code Chat interface, with Copilot automatically accessing necessary context, creating GitHub issues, assigning tasks, and monitoring progress through a dedicated Agent session dashboard. **Key Points:** - GitHub Copilot Agents were demonstrated handling complex tasks like adding accessibility features (high contrast mode) and implementing fuzzy search functionality for the Tailspin Toys application. - The process begins in the VS Code Chat interface, where a user inputs a task, potentially selects a specific custom agent (like the 'Search engine optimization (SEO)' agent), and initiates the task. - For the high contrast mode task, Copilot automatically created a GitHub issue (#56), assigned itself, and developed an implementation plan detailing CSS changes, toggle controls, and testing. - For the search feature, Copilot used an existing SEO agent playbook, performed file discovery (listing 11 files in server/routes), implemented fuzzy SQL LIKE queries, and created 9 Playwright E2E tests. - The workflow is managed via the Agents dashboard, showing session status (e.g., 'In progress', 'Ready for review'), and allows users to monitor progress or steer the active session. - The video also showed how Copilot can interact with Slack, responding to a direct request to update the site to a Progressive Web App (PWA) by creating a new branch and pull request on GitHub. - The demonstration highlights the power of agents to work autonomously across repositories, GitHub, and external tools like Slack to achieve substantial development goals. ![Screenshot at 0:07: The initial prompt in the VS Code Chat interface asking Copilot to provide actionable bullet points for accessibility requirements, setting the stage for agent interaction.](https://ss.rapidrecap.app/screens/__RLpdxThEE/00-00-07.png) **Context:** The video showcases the advanced capabilities of GitHub Copilot Agents, which extend the functionality of Copilot beyond simple code completion to manage complex, multi-stage development workflows. The demonstration centers on a sample application, 'Tailspin Toys,' where the user interacts with the Agents interface within VS Code Chat to delegate tasks related to accessibility and feature implementation, such as adding high contrast modes and search functionality. ## Detailed Analysis The demonstration illustrates how GitHub Copilot Agents streamline development by handling complex tasks through natural language commands within the VS Code Chat environment. Initially, the user requests accessibility guidance for the Tailspin Toys app, leading Copilot to use context-aware tools to list available Copilot Spaces and access the 'Accessibility' space for guidance, resulting in a list of actionable bullet points covering testing, semantic HTML, and media handling. Subsequently, a request is made to implement a high contrast mode, which triggers Copilot to create and assign itself a GitHub Issue (#56), detailing an implementation plan including CSS custom properties, toggle controls, and E2E testing via Playwright. The video then shifts focus to search functionality, where the user selects the custom 'Search engine optimization (SEO)' agent, which utilizes its predefined playbook to implement fuzzy search logic in the backend (server/routes/games.py) using SQL's ILIKE operator, and creates corresponding Playwright and unit tests. Finally, the workflow transitions to Slack, where a user requests a PWA update, prompting Copilot to create a new branch ('copilot/update-site-to-pwa') and open a Pull Request on GitHub, confirming the task is in progress. The Agents dashboard provides a central view of all ongoing tasks, indicating their status, progress, and resource usage, confirming that agents can manage tasks end-to-end, from initial request to code changes and integration. ### Accessibility Task Delegation - User requests accessibility guidance - Copilot lists available Copilot Spaces - Copilot accesses the 'Accessibility' space for guidance - Initial actionable bullet points are generated. ### High Contrast Mode Implementation - User requests a high contrast mode toggle feature - Copilot creates and assigns GitHub Issue #56 - Implementation plan details CSS, header controls, local storage persistence, and testing. ### Fuzzy Search Feature Implementation - User selects the 'Search engine optimization (SEO)' agent - Copilot lists relevant files in server/routes and server/models - Implements fuzzy search using SQL ILIKE in games.py - Creates 9 Playwright E2E tests. ### Agent Workflow Management - Agents dashboard displays session status (e.g., 'In progress', 'Ready for review') - Sessions list shows progress metrics (+687, -23) for completed tasks. ### Integration with External Tools (Slack/GitHub) - User requests PWA update in Slack - Copilot responds by creating a GitHub PR (#38) with branch 'copilot/update-site-to-pwa' - Copilot confirms app installation and begins work. ### SEO Agent Playbook Review - Viewing the 'seo-agent.md' file reveals core principles like focusing on user intent, metadata guidelines, and technical guardrails for Copilot to follow when optimizing content. ![Screenshot at 0:07: The initial prompt in the VS Code Chat interface asking Copilot to provide actionable bullet points for accessibility requirements, setting the stage for agent interaction.](https://ss.rapidrecap.app/screens/__RLpdxThEE/00-00-07.png) ![Screenshot at 0:13: Copilot successfully listing available GitHub Copilot Spaces, indicating context gathering for the accessibility task.](https://ss.rapidrecap.app/screens/__RLpdxThEE/00-00-13.png) ![Screenshot at 0:23: The generated list of 'Quick Accessibility Action Items' including running automated tests with axe-core/playwright and testing keyboard navigation.](https://ss.rapidrecap.app/screens/__RLpdxThEE/00-00-23.png) ![Screenshot at 0:48: The user assigning a specific task to Copilot via the chat interface to create a high contrast mode feature, which triggers the agent workflow.](https://ss.rapidrecap.app/screens/__RLpdxThEE/00-00-48.png) ![Screenshot at 1:14: The GitHub Copilot CLI launches in the terminal after using the 'copilot --allow-all-tools' command, showing connectivity to the MCP Server.](https://ss.rapidrecap.app/screens/__RLpdxThEE/00-01-14.png) ![Screenshot at 1:24: Copilot detailing the plan for adding fuzzy search, including listing files in server/routes and server/models directories for modification.](https://ss.rapidrecap.app/screens/__RLpdxThEE/00-01-24.png) ![Screenshot at 2:12: The resulting GitHub Pull Request \(#38\) detailing the 'Update site to support Progressive Web App features' task, showing Copilot's initial comment summarizing the plan.](https://ss.rapidrecap.app/screens/__RLpdxThEE/00-02-12.png) ![Screenshot at 3:09: The final review screen in the Agent interface showing the completed Pull Request summary, including 'Exceeds Requirements' and detailed changes for the theme toggle feature.](https://ss.rapidrecap.app/screens/__RLpdxThEE/00-03-09.png) ![Screenshot at 3:36: Viewing the 'seo-agent.md' file in GitHub, which contains the specific SEO playbook principles used by the dedicated agent.](https://ss.rapidrecap.app/screens/__RLpdxThEE/00-03-36.png)