Skip to content
| Marketplace
Sign in
Visual Studio Code>Machine Learning>ACE for GitHub CopilotNew to Visual Studio Code? Get it now.
ACE for GitHub Copilot

ACE for GitHub Copilot

code engine GmbH

| (1) | Free
Pattern learning and retrieval for GitHub Copilot - captures what works, retrieves it when needed
Installation
Launch VS Code Quick Open (Ctrl+P), paste the following command, and press enter.
Copied to clipboard
More Info

ACE for VS Code

🔐 CLOSED BETA - This extension requires an ACE account to work.

→ Join the Waitlist to get your API token and start using ACE.

Without an account, the extension cannot connect to the ACE server.

Automatic Context Engine - Pattern learning for GitHub Copilot that captures what works and retrieves it when needed.

By Code Engine GmbH

Features

  • 🤖 @ace Chat Participant - Ask Copilot to search patterns, capture learning, and more
  • 🔧 Language Model Tools - ace_search, ace_learn, ace_status, ace_get_playbook
  • 📁 Multi-Root Workspace Support - Each folder gets its own ACE configuration
  • 📊 Status Panel - Real-time playbook statistics
  • ⚙️ Configure Panel - Easy server and project setup
  • 🤖 Agent Files - Generate .github/agents/ for CI/CD AI assistants

Quick Start

Step 1 is required - The extension won't work without an ACE account!

  1. Sign up for ACE - Get your API token (required!)
  2. Install from VS Code Extensions marketplace
  3. Open Command Palette (Cmd+Shift+P on Mac, Ctrl+Shift+P on Windows/Linux)
  4. Run ACE: Configure - opens setup panel
  5. Enter your credentials and select your project
  6. Run ACE: Update Agent Files - generates .github/agents/ for CI/CD AI assistants
  7. Start using @ace in GitHub Copilot Chat!

Using ACE with Copilot

Agent Mode (Recommended)

For automatic pattern retrieval before tasks, select the ace-expert agent in Copilot Chat:

  1. Open GitHub Copilot Chat (Cmd+I or Ctrl+I)
  2. Click the agent selector dropdown (top of chat panel)
  3. Select ace-expert from the list
  4. Start coding - ACE tools are now available to Copilot automatically!

When ace-expert is selected, Copilot can automatically call ace_search before implementing and ace_learn after completing work.

Chat Commands

Alternatively, use @ace commands directly in any Copilot chat:

@ace Chat Participant

Use the @ace chat participant in GitHub Copilot Chat:

Command Description
@ace /search <query> Search for relevant patterns in the playbook
@ace /patterns [section] View playbook patterns by section
@ace /status Show playbook statistics
@ace /learn Capture patterns from the current session
@ace /top Show highest-rated patterns
@ace /bootstrap Initialize playbook from codebase
@ace /clear --confirm Clear all patterns

Language Model Tools

ACE provides 4 tools that GitHub Copilot can use automatically:

Tool Description
ace_search Search playbook for relevant patterns before implementing
ace_learn Capture patterns and lessons learned from work
ace_status Show playbook statistics
ace_get_playbook Retrieve patterns filtered by section or quality

Commands

Command Description
ACE: Configure Opens webview panel to configure server URL, API token, and select project
ACE: Show Playbook Status Opens webview panel showing playbook statistics
ACE: Bootstrap Playbook Initialize patterns from existing codebase (git history, docs)
ACE: Capture Learning Manually trigger learning capture
ACE: Clear Playbook Clear all learned patterns
ACE: Update Agent Files Generate/update .github/agents/ files
ACE: Quick Actions Quick access to common ACE actions

How It Works

ACE creates a self-improving learning cycle where each session benefits from previous work:

Step Action Description
🚀 Start Task You begin work in VS Code
⬇️
📖 Retrieve Patterns Copilot calls ace_search → Fetches learned strategies
⬇️
⚡ Execute Copilot completes task using patterns
⬇️
💡 Capture Learning Copilot calls ace_learn → Playbook grows smarter
⬇️
🔄 Next Session Enhanced patterns available → Back to Start!

Playbook Sections

  1. strategies_and_hard_rules - Architectural patterns, coding principles
  2. useful_code_snippets - Reusable code patterns with context
  3. troubleshooting_and_pitfalls - Known issues, gotchas, solutions
  4. apis_to_use - Recommended libraries, frameworks, integration patterns

Multi-Root Workspace Support

In multi-root workspaces, each folder can have its own ACE configuration:

  • Config panel shows folder name in title and header
  • Per-folder settings saved to .vscode/settings.json
  • Folder picker appears when multiple folders are open
  • All chat commands are folder-aware

Configuration

Workspace Settings

Configure in VS Code Settings (Cmd+,) or .vscode/settings.json:

Setting Description Default
ace.projectId ACE Project ID (format: prj_xxxxx) ""
ace.orgId ACE Organization ID (format: org_xxxxx) ""
ace.serverUrl ACE API Server URL https://ace-api.code-engine.app
ace.automation.level Automation level: manual, smart, aggressive smart
ace.automation.minEditsBeforeSuggest Minimum edits before suggesting learning 10
ace.automation.idleMinutesBeforeSuggest Idle minutes before suggesting learning 3
ace.automation.showStatusBar Show ACE status in status bar true
ace.automation.gitAutoCapture Auto-suggest learning on commits false

Agent Files

ACE can generate .github/agents/ files for CI/CD AI assistants. Run ACE: Update Agent Files to create:

  • .github/agents/ace-playbook.md - Playbook patterns for AI agents
  • .github/.ace-version.json - Version tracking for smart updates

Requirements

  • VS Code (v1.102.0+)
  • GitHub Copilot Chat extension
  • ACE account at ace.code-engine.app
  • API token from the ACE dashboard

Links

  • ACE Website
  • Documentation
  • GitHub
  • Report Issues

License

MIT - Code Engine GmbH

  • Contact us
  • Jobs
  • Privacy
  • Manage cookies
  • Terms of use
  • Trademarks
© 2025 Microsoft