685.4-iteration Iterate on CLI and plugin based on testing feedback

[codekiln]

Parent Issue

Sub-issue of #685 (ls-claude-code-plugin milestone)

Phase

0.4: Iteration

Purpose

Based on QA findings from phase 0.3, improve both the CLI (via ls-prompt-ux) and plugin commands. This is the co-evolution feedback loop in action.

Scope

1. Address CLI Issues

For each CLI pain point identified:

• File issue or comment in ls-prompt-ux milestone (🔧 build: update Claude Code plugin configuration #16)
• Track CLI improvements
• Update plugin when CLI changes

2. Improve Plugin Commands

Based on testing:

• Refine instructions for clarity
• Add missing error handling guidance
• Adjust allowed-tools scope if needed
• Improve context section based on what Claude needs

3. Add Sub-Agents (if needed)

If context management is a problem:

• Implement prompt-finder sub-agent
• Test with haiku model for filtering
• Document context savings

4. Document Learnings

Create guidelines:

• "What makes CLI output Claude-friendly?"
• "Effective slash command patterns"
• "When to use sub-agents"

Deliverables

• Updated /langstar-prompt-workshop command
• Issues/comments filed in ls-prompt-ux (🔧 build: update Claude Code plugin configuration #16)
• Sub-agents if needed
• Guidelines document

Dependencies

• 685.3-qa-framework Establish QA methodology for agentic coding instructions #688 (QA framework) provides test results
• ls-prompt-ux milestone (🔧 build: update Claude Code plugin configuration #16) for CLI changes

Success Criteria

• All critical QA findings addressed
• Plugin performs noticeably better after iteration
• Guidelines documented for future development