Clarification on GitHub Copilot Skills, GitHub Pro, and a private repository-based memory workflow #210552
Replies: 4 comments
|
Hi @xiangjishi1983-commits! Here is an official product-level breakdown answering your four questions regarding Copilot Skills, GitHub Pro, and private repository workflows: 1. GitHub Pro vs. GitHub CopilotGitHub Pro and GitHub Copilot are distinct product subscriptions:
2. Runtime for the
|
| Component | Provided by GitHub / Copilot | User-Managed |
|---|---|---|
| Language Model & Inference | ✅ GitHub Copilot API | ❌ |
| Skill Instructions | ✅ Pre-packaged prompts | ❌ |
| Agent Execution Runtime | ✅ VS Code / JetBrains extension | Custom daemon if running headless |
| Memory Persistence | ❌ None (stateless across sessions) | ✅ Custom repo storage / vector index |
| Approval Gate | ✅ Native PR review workflows | Workflow triggers |
This comment was marked as low quality.
This comment was marked as low quality.
|
GitHub Pro manages account features but lacks AI capabilities, which require a Copilot subscription and organization policies. Copilot skills are static instruction bundles executed through supported clients like VS Code or the Copilot CLI without requiring external microservices. GitHub Actions provides a hosted environment for validation, while GitHub ensures strict repository isolation and protects private code from training global models. Ultimately, GitHub supplies the platform features, runtime interfaces, memory mechanisms, and CI infrastructure, whereas users are responsible for authoring skill content, configuring custom workflows, setting repository permissions, and managing validation rules. |

Uh oh!
There was an error while loading. Please reload this page.
🏷️ Discussion Type
Question
💬 Feature/Topic Area
Copilot in GitHub
Body
Hello GitHub Community,
I’m looking for clarification on what GitHub officially supports for a private-repository-based memory workflow using GitHub Copilot Skills. The goal is to generate reviewable, project-scoped memory candidates from user-authorized context while keeping persistence and approval under explicit control. I’m asking about product capabilities and will not include private repository contents.
GitHub Pro: Does GitHub Pro (distinct from GitHub Copilot Pro) provide capabilities relevant to this workflow, such as private-repository access for Copilot Skills, persistent project context, or GitHub-hosted validation? If not, which plan or product provides them?
Runtime for the
rememberSkill: Does GitHub’srememberSkill require a separately hosted agent/runtime, such as VS Code or another client, or can it run in a GitHub-hosted Copilot experience? Is a Skill itself only an instruction bundle that must be loaded by a compatible agent?GitHub-hosted validation: Is there an official GitHub-hosted way to validate a Skill-driven workflow against a private repository (for example, GitHub Actions, Copilot coding agent, or another hosted execution environment) without deploying an external compute service? What security boundaries apply?
What users must integrate: Which pieces must users build or configure themselves—Skill installation/loading, agent runtime, repository permissions, persistent storage or memory backend, approval gates, CI/validation, and project isolation? Which of these, if any, are provided automatically by GitHub or Copilot?
Pointers to current documentation and applicable plan requirements would be appreciated. I understand that an instruction-only Skill does not itself grant permissions or provide persistence; I’m asking which supported products or services are available and what must be integrated separately.
All reactions