Team Patterns

This page covers configuration deep-dives, troubleshooting, team standardization, and performance optimization. It wraps up the workshop with a summary of everything you've learned.

For autopilot, fleet, and LSP configuration, see Automation Patterns. For foundational configuration and CI/CD, see Advanced Topics.

Exercise 8: Configuration File Deep Dive

Goal: Understand and customize config.json and related configuration files.

Steps:

1. View current configuration:

   cat ~/.copilot/config.json

2. Example full configuration:

   {
     "trusted_folders": [
       "/home/user/projects",
       "/home/user/work"
     ],
     "default_model": "gpt-4.1",
     "theme": "dark",
     "editor": "code",
     "shell": "bash",
     "telemetry": true,
     "auto_update": true
   }

3. Add trusted folders:

   # Using jq to update config
   jq '.trusted_folders += ["/new/path"]' ~/.copilot/config.json > tmp.json
   mv tmp.json ~/.copilot/config.json

4. Configure URL restrictions:

   {
     "web_fetch": {
       "allowed_urls": [
         "https://api.github.com/*",
         "https://docs.github.com/*"
       ],
       "denied_urls": [
         "https://evil.com/*"
       ]
     }
   }

Expected Outcome: Custom configuration for your workflow, including LSP settings from Exercise 7.

Exercise 9: Troubleshooting Guide

Goal: Diagnose and fix common issues.

Steps:

1. Authentication issues:

   # Clear credentials and re-authenticate
   rm -rf ~/.copilot/auth*
   copilot
   # Follow OAuth flow

2. Tool not working:

   # Check if tool is allowed
   copilot -p "test" --available-tools
   
   # Verify MCP servers
   copilot
   /mcp show

3. Session issues:

   # Clear session data
   rm -rf ~/.copilot/sessions/
   
   # Start fresh
   copilot
   /clear

4. Performance issues:

   # Check context usage
   copilot
   /context
   
   # Compact if needed
   /compact
   
   # Or start fresh
   /clear

5. Agent not found:

   # Verify agent file exists
   ls -la .github/agents/
   
   # Check YAML syntax
   cat .github/agents/my-agent.md | head -10
   
   # Ensure frontmatter is valid

6. MCP server failing:

   # Check if server runs independently
   npx @modelcontextprotocol/server-memory
   
   # Check config syntax
   cat ~/.copilot/mcp-config.json | jq .

Expected Outcome: You can diagnose and resolve common problems, including LSP timeouts and shell mode access.

Exercise 10: Team Workflow Patterns

Goal: Establish team-wide Copilot practices.

Steps:

1. Standardize repository configuration:

   # Create a template repository with:
   .github/
   ├── copilot-instructions.md    # Team coding standards
   ├── agents/
   │   ├── reviewer.md            # Code review agent
   │   └── docs.md                # Documentation agent
   ├── instructions/
   │   ├── typescript.instructions.md
   │   └── tests.instructions.md
   └── hooks/
       └── hooks.json             # Security guardrails
   
   AGENTS.md                      # Project-specific agent
   llm.txt                        # Project context for LLMs

2. Create onboarding documentation:

   # Team Copilot CLI Guide
   
   ## Setup
   1. Install: `npm install -g @github/copilot`
   2. Authenticate: `copilot` and follow OAuth
   3. Clone this repo template
   
   ## Our Agents
   - @reviewer - Code review
   - @docs - Documentation
   
   ## Commit Convention
   All commits must follow Conventional Commits.
   
   ## Security Rules
   - Never approve `rm -rf` for session
   - Never push directly to main
   - Always use `--deny-tool 'shell(git push)'` in automation

3. Share MCP configurations:

   # Create a shared MCP config
   cat > shared-mcp-config.json << 'EOF'
   {
     "servers": {
       "team-tools": {
         "url": "https://team-mcp.example.com/",
         "requestInit": {
           "headers": {
             "Authorization": "Bearer ${TEAM_MCP_TOKEN}"
           }
         }
       }
     }
   }
   EOF

4. Establish review checklist:

   ## PR Copilot Checklist
   - [ ] Run `@reviewer` on changes
   - [ ] Generate docs with `@docs`
   - [ ] Check context usage stayed reasonable
   - [ ] No sensitive data in session exports

Expected Outcome: Team-wide standardization on Copilot usage, including shared LSP and environment configurations.

Exercise 11: Performance Optimization

Goal: Get the best performance from Copilot CLI.

Steps:

1. Optimize startup time:

   # Pre-authenticate
   copilot --version
   
   # Use --silent for faster scripted operations
   copilot -p "quick task" --silent

2. Reduce network round-trips:

   # Batch operations
   copilot -p "Create user.ts, user.test.ts, and user.types.ts with related code" \
     --allow-tool 'write'
   
   # Instead of three separate prompts

3. Choose appropriate models:

   # Fast model for simple tasks
   copilot --model gpt-5-mini -p "Format this JSON"
   
   # Full model for complex analysis
   copilot --model gpt-4.1 -p "Refactor this complex module"

4. Efficient context management:

   # Use @explore for overview without context cost
   # Use targeted reads instead of reading all files
   # Compact proactively, not reactively

5. Use Autopilot for repetitive multi-step tasks:

   # Instead of manual step-by-step
   copilot --autopilot -p "Set up complete testing infrastructure for all services"
   
   # Saves time and reduces back-and-forth

   See Exercise 4 for full autopilot details.

6. Use Fleet for parallel work:

   # Faster than sequential processing
   copilot
   /fleet "Migrate all 50 components to the new API format"
   
   # Multiple agents work in parallel

   See Exercise 5 for full fleet details.

7. Parallel sessions for independent tasks:

   # Terminal 1: Frontend work
   cd frontend && copilot
   
   # Terminal 2: Backend work
   cd backend && copilot
   
   # Terminal 3: Documentation
   cd docs && copilot

Expected Outcome: Maximum performance from Copilot CLI using parallelization, autopilot, and fleet features.

Summary

• ✅ Environment variables customize behavior globally
• ✅ CI/CD integration enables automated workflows
• ✅ Advanced flags give precise control
• ✅ Autopilot mode enables autonomous multi-step task execution (v0.0.411)
• ✅ Fleet command parallelizes complex tasks with orchestrator validation (v0.0.411-v0.0.412)
• ✅ --bash-env sources custom environment for shell sessions (v0.0.412)
• ✅ --experimental enables alt-screen mode by default (v0.0.413)
• ✅ LSP configuration controls language server timeouts; default request timeout is 90s (v0.0.412-v0.0.413)
• ✅ Shell mode access via ! command (v0.0.410)
• ✅ config.json and lsp.json persist preferences
• ✅ Team standardization ensures consistency
• ✅ Performance optimization maximizes productivity

Workshop Complete! 🎉

Congratulations on completing the GitHub Copilot CLI Workshop!

What You've Learned

1. Installation - Multiple methods to install and authenticate
2. Operating Modes - Interactive, programmatic, and delegate
3. Sessions - Management, persistence, and control
4. Instructions - AGENTS.md, copilot-instructions.md, llm.txt
5. Tools - Permissions, allow/deny, YOLO mode
6. MCP Servers - Configuration and custom integrations
7. Skills - Creating and using specialized capabilities
8. Plugins - Ecosystem and custom extensions
9. Custom Agents - Building specialized personas
10. Hooks - Lifecycle automation and security
11. Context - Monitoring and optimization
12. Advanced - Autopilot, Fleet, CI/CD, LSP config, environment, and team practices

Next Steps

• Practice daily with real projects
• Create custom agents for your workflow
• Share skills with your team
• Contribute to the Copilot ecosystem

Resources

• GitHub Copilot Documentation
• Copilot CLI Blog Posts
• GitHub Community Discussions
• agentskills.io

---

Thank you for participating in this workshop!