fix(skills): use chained commands in quickstart skill

[adnaan]

Summary

• Fix quickstart skill to use chained commands with && instead of separate lines
• Align quickstart skill format with the plan skill for consistency
• Add explicit instruction to NOT execute commands one-by-one iteratively
• Reduce file size from ~450 lines to ~145 lines by removing redundant content

Problem

The quickstart skill was showing commands as separate lines:

lvt new [name] --kit multi
cd [name]
lvt gen resource [resource] [fields...]
lvt migration up
go mod tidy

This caused Claude to execute commands iteratively, waiting for output after each command instead of running them as a single script.

Solution

Updated to use chained commands matching the plan skill:

lvt new ___NAME___ --kit multi && \
  cd ___NAME___ && \
  lvt gen resource ___RESOURCE___ ___FIELDS___ && \
  lvt gen auth && \
  lvt migration up && \
  go mod tidy && \
  lvt seed ___RESOURCE___ --count 50

Test plan

• Verify quickstart skill presents chained command format
• Verify "yes" approval executes single chained command
• Verify plan and quickstart skills produce consistent output format

🤖 Generated with Claude Code

[claude]

PR Review: fix(skills): use chained commands in quickstart skill

Summary

This PR significantly improves the quickstart skill by adopting chained commands (using &&) instead of separate command execution, aligning it with the plan skill's format. The refactor reduces the file from ~450 lines to ~145 lines while making the skill more reliable and consistent.

✅ Strengths

1. Excellent Consistency: The quickstart skill now matches the plan skill's format exactly, which improves maintainability and reduces cognitive load for users who interact with both skills.

2. Better Execution Model: Using chained commands with && ensures:

   • All commands execute as a single atomic operation
   • Automatic failure handling (stops on first error)
   • Prevents iterative execution that was causing the original issue

3. Reduced Redundancy: Cutting ~300 lines of redundant step-by-step instructions improves:

   • File readability
   • Maintenance burden
   • Loading/parsing time
   • Focus on essential content

4. Clear Structure: The new template-based approach with fill-in-the-blanks is much clearer:

   • Domain detection table is explicit and easy to understand
   • Command templates are well-separated (with auth vs without auth)
   • Rules are concise and unambiguous

5. Proper Version Bump: Bumping from 1.1.0 to 2.0.0 correctly signals a breaking behavior change.

🔍 Code Quality

Excellent

• Clear, imperative instructions for Claude
• Well-structured sections with proper Markdown hierarchy
• Good use of examples and formatting
• Consistent with repository conventions (based on plan skill)

🐛 Potential Issues

Minor Concerns:

1. Missing Instruction Clarity (Line 80): The "DO NOT ask questions" instruction might be too strong. Consider:

   • What if the user request is genuinely ambiguous? (e.g., "create an app" with no domain)
   • Recommend adding: "DO NOT ask questions UNLESS the user request lacks essential information (app type/domain)"

2. Domain Detection Edge Cases (Lines 38-46):

   • The "(other)" fallback uses generic "app/items" which may not be intuitive
   • Consider adding guidance for common terms like "inventory", "booking", "event", etc.

3. Error Handling: The chained command using && will stop on first failure, but there's no explicit guidance on:

   • How to resume after partial success
   • Whether to clean up partially created projects
   • Suggest adding a troubleshooting note

4. Command Validation (Lines 51-70): The templates hardcode --kit multi and --count 50. While these are sensible defaults, consider documenting why these specific values were chosen.

🚀 Performance Considerations

Positive Impact:

• Reduced file size (450 → 145 lines) improves parsing speed
• Single chained command execution is more efficient than multiple round-trips
• Reduced context reduces token usage in LLM interactions

No Performance Concerns Identified

🔒 Security Concerns

No Security Issues

The changes don't introduce any security vulnerabilities:

• Commands are properly chained with &&
• No user input is directly interpolated without context
• Template placeholders (NAME, etc.) follow safe substitution patterns

🧪 Test Coverage

Test Plan Provided: The PR includes a test plan with 3 items:

• ✅ Verify quickstart skill presents chained command format
• ✅ Verify "yes" approval executes single chained command
• ✅ Verify plan and quickstart skills produce consistent output format

Recommendation: Consider adding automated tests for:

1. Domain detection logic (all 6 domain cases)
2. Field parsing and validation
3. Template substitution correctness
4. Command generation for both auth and non-auth scenarios

📝 Suggestions for Improvement

1. Add Resumption Guidance:

## IF COMMAND FAILS

If the chained command fails partway through:
1. Check which step failed: `cd [name] && pwd` (did directory get created?)
2. If app was created, cd into it and resume: `lvt gen resource...`
3. If app creation failed, clean up and retry: `rm -rf [name] && [full command]`

2. Enhance Domain Detection Table:

| inventory/warehouse | inventory | inventory | items | name:string quantity:int location:string | Password |
| booking/reservation | booking | booking | reservations | name:string date:time status:string | Password |
| event/calendar | events | events | events | title:string date:time location:string | Password |

3. Document Default Choices:

## DEFAULTS RATIONALE
- Kit: `multi` (full-page layout is most versatile for new apps)
- Seed count: 50 (enough to test pagination, not overwhelming)
- Auth: Password (most commonly needed, easiest to set up)

4. Add Success Indicators:

After successful execution, user should see:
- ✅ "Database migrations applied"
- ✅ "50 [resource] records seeded"
- ✅ "Ready to serve on :8080"

🎯 Overall Assessment

Rating: ⭐⭐⭐⭐⭐ (5/5)

This is an excellent improvement that:

• Solves a real user experience problem (iterative vs atomic execution)
• Significantly improves code quality and maintainability
• Maintains backward compatibility through proper versioning
• Aligns with established patterns in the codebase

Recommendation: Approve and merge with consideration for the minor suggestions above in a follow-up PR.

✨ Excellent Work!

The refactor is well-thought-out and the alignment with the plan skill will make both skills easier to maintain going forward. The reduction in file size without loss of functionality is impressive.

---

Reviewed by: Claude Sonnet 4.5 🤖

[Copilot]

Pull request overview

This PR refactors the quickstart skill to use chained commands (with &&) instead of separate command lines, aligning it with the plan skill format. The goal is to prevent Claude from executing commands iteratively and instead run them as a single chained script.

Changes:

• Updated quickstart skill to present commands as a single chained command with && operators
• Added explicit instruction to NOT execute commands one-by-one iteratively
• Removed ~300 lines of redundant documentation and examples while preserving core functionality

[Copilot]

The horizontal rule separator uses inconsistent syntax. In markdown, horizontal rules should use three or more hyphens without the 0: prefix. The correct format should be --- on its own line.

[Copilot]

The template shows ___FIELDS___ as a single placeholder, but the FILL IN BLANKS table shows fields like title:string content:text published:bool. The documentation should clarify that multiple space-separated field definitions should replace ___FIELDS___, not just a single value. This could confuse users about the expected format.

[Copilot]

Corrected capitalization inconsistency. The instruction uses 'Do NOT' while other similar instructions use 'DO NOT' (all caps).