feat(ai): make AI tool configs first-class, consistent, and profile-aware

[Jake-Prickett]

Summary

• Copilot instructions → template: global-copilot-instructions.md renamed to .md.tmpl; adds a ## Personal Profile Context section (iOS/SwiftUI, React Native/Expo, Next.js) gated on personal profile, plus a rendered OS/profile footer
• VS Code MCP config → template: mcp.json renamed to mcp.json.tmpl; entire file wrapped in darwin OS guard; Figma and Notion entries gated on personal profile; GitHub MCP unconditional
• MCP health check: validate-setup.sh gains check_mcp_server() — warns if github-mcp-server is missing from PATH or gh is unauthenticated
• AGENTS.md memory note: Added ### Memory subsection under Claude Code in AGENTS.md.tmpl documenting the persistent memory path
• Docs refresh: docs/ai-tools-integration.md updated with MCP servers table (Tool/Server/Transport/Auth), Claude commands table, Codex skills table (all 9 skills), and a project bootstrapping step-by-step
• AGENTS-TEMPLATE.md: New static starter at private_dot_claude/AGENTS-TEMPLATE.md for seeding project-level AGENTS.md files

Test plan

• chezmoi diff — confirm rendered changes look correct for both personal and work profiles
• chezmoi apply — apply to home directory
• ./scripts/validate-setup.sh — confirm MCP health check passes (or warns correctly if unauthenticated)
• Verify ~/.config/github-copilot/global-copilot-instructions.md renders with profile section on personal, without on work
• Verify ~/Library/Application Support/Code - Insiders/User/mcp.json renders with only GitHub MCP on work profile

🤖 Generated with Claude Code

[claude]

Claude finished @Jake-Prickett's task —— View job

---

Code Review Complete ✅

I've analyzed this AI tooling improvements PR and overall it's a well-executed enhancement that makes the AI tool configurations more maintainable and profile-aware. Here's my detailed review:

• Read and analyze all changed files
• Review chezmoi template syntax and logic
• Check profile-aware implementation
• Evaluate security considerations
• Assess documentation quality
• Review validation approach and test coverage
• Provide structured feedback

🟢 Strengths

1. Excellent Template Design

• Proper Guards: All profile and OS guards use correct chezmoi template syntax ({{ if eq .profile "personal" }}, {{ if eq .chezmoi.os "darwin" }})
• Clean Logic: Profile-specific content is properly gated without creating empty sections
• Consistent Patterns: Template structure follows established conventions across all files

2. Strong Security Implementation

• No Hardcoded Secrets: GitHub MCP correctly uses $(gh auth token) at runtime
• Proper Permissions: Maintains private_ prefix for restricted files
• Environment Variables: API keys properly sourced from env vars, not hardcoded

3. Comprehensive Documentation

The docs/ai-tools-integration.md update is particularly well done:

• Clear MCP servers table with transport and auth details
• Complete tool coverage (Claude Code, Codex, VS Code)
• Helpful project bootstrapping steps

4. Thoughtful Validation

The MCP health check addition in validate-setup.sh:72-83 is excellent:

• Checks both github-mcp-server availability and gh authentication
• Provides clear, actionable warning messages
• Follows the script's existing pattern

🟡 Minor Improvements

1. Template Syntax Consistency (Line-specific feedback)

File: private_dot_config/github-copilot/global-copilot-instructions.md.tmpl:42-48

{{ if eq .profile "personal" }}
## Personal Profile Context

Primary stacks: iOS/SwiftUI, React Native/Expo, Next.js
Prefer Swift concurrency (async/await, actors) over GCD.
Expo Router for navigation in React Native projects.
{{ end }}

Suggestion: Consider adding a blank line before the closing {{ end }} for better readability when rendered.

2. VS Code MCP Configuration

File: private_Library/.../mcp.json.tmpl:4-16
The profile guard for Figma and Notion MCPs is well-implemented, but consider adding version validation or fallback handling if the gallery API changes.

3. Documentation Enhancement

File: docs/ai-tools-integration.md:123-131
The project bootstrapping section could benefit from numbering the CLAUDE-TEMPLATE.md.tmpl reference (should be CLAUDE.md.tmpl based on the context).

🟢 Test Coverage Assessment

The PR includes solid validation through:

• validate-setup.sh MCP health check function
• Clear test plan in PR description
• Profile-aware rendering verification steps

Recommended test additions:

1. Verify template rendering on both personal and work profiles
2. Test MCP server connectivity after applying changes
3. Confirm VS Code MCP config loads correctly on macOS

🟢 Performance & Architecture

• Minimal Runtime Impact: Templates only execute during chezmoi apply
• Clean Separation: Profile logic doesn't create complex conditionals
• Maintainable Structure: Easy to add new AI tools following established patterns

🔒 Security Review

✅ Passed all security checks:

• No secrets in tracked files
• Proper use of environment variables
• Runtime token acquisition vs. storage
• Appropriate file permissions with private_ prefix

📋 Final Recommendations

1. Ready to Merge: This PR is well-implemented and follows best practices
2. Test the Test Plan: Run through the verification steps in the PR description
3. Consider Future Enhancements: The architecture supports easy addition of new AI tools

🎯 Code Quality Score: 9/10

This is high-quality infrastructure code that makes AI tooling configuration more maintainable and secure. The template design is clean, security practices are solid, and documentation is comprehensive.

---