Glossary¶
Look up a term here, then follow its session link for an explanation and practice. Each link points to the first session that teaches the topic.
Agent mode¶
An IDE workflow where Copilot can plan, edit files, run tools, and respond to results. See Session M2-02.
Agent skill¶
A reusable SKILL.md package that gives an agent task-specific instructions and
supporting resources. See Session M3-05.
Automation¶
A reusable Copilot task that runs manually or on a schedule. See Session M7-06.
Canvas¶
An interactive view in the Copilot App. People and agents work with the same state through its actions. See Session M3-01 and Session M7-04.
Cloud agent¶
An asynchronous Copilot agent that works in a GitHub-hosted environment and returns changes for human review. See Session M3-02.
Copilot App¶
GitHub's desktop application for agent sessions, projects, parallel work, and review. See Session M7-01 and Session M2-05.
Custom agent¶
A specialized agent profile with its own instructions, tools, and access boundaries. See Session M3-04.
Custom instructions¶
Repository, path, organization, or personal guidance that shapes how Copilot works with a codebase. See Session M2-03.
GitHub MCP server¶
The Model Context Protocol server that exposes approved GitHub reads and writes to Copilot. See Session M3-03 and Session M7-03.
Human gate¶
A required review or approval before an agent can continue or its work can be accepted. See Session M6-02.
Model Context Protocol (MCP)¶
A standard interface for connecting an agent to external tools and data. See Session M3-03.
Pull request readiness¶
Having enough information about the scope, test results, review findings, and remaining risks for a person to decide whether the change is ready for a pull request. See Session M2-05.
RPI¶
The HVE Research, Plan, Implement, Review, and Follow-up delivery loop. See Session M5-03.
Space¶
A set of repositories and other context you choose to guide Copilot's responses. See Session M2-03.
Specification-driven development¶
A workflow where you review requirements and plans before implementation, then keep them aligned with the code and tests. See Session M5-02.