presonus-studiolive-mcp: Command for GitHub Copilot

Requirements Prompts Improvements

Date: November 6, 2025
Issue: Copilot not consistently delivering expected outputs when executing requirements prompts
Solution: Added explicit "Expected Output" sections with complete templates

---

🔍 Root Cause Analysis

Problems Identified

1. No Clear Output Format at Top

   • Deliverable templates were buried deep in the prompt
   • Copilot had to read extensive guidance before understanding what to produce
   • Result: Sometimes stopped after questions, didn't generate final output

2. Unclear Workflow

   • Missing "START HERE → DO THIS → DELIVER THAT" structure
   • No explicit steps showing what to do when
   • Result: Copilot unclear about sequence of actions

3. Example-Heavy Content

   • Many examples but unclear which was the actual deliverable
   • Guidance overwhelmed the task definition
   • Result: Copilot focused on examples rather than producing output

4. Missing Explicit Instructions

   • No "ALWAYS DELIVER" or "MUST PRODUCE" language
   • Lacked emphasis on completing the full deliverable
   • Result: Partial outputs or summaries instead of complete documents

---

✅ Solutions Implemented

1. Added "📤 EXPECTED OUTPUT" Section

Location: Immediately after the agent role definition (lines 10-80 typically)

Format:

## 📤 EXPECTED OUTPUT (ALWAYS DELIVER)

When [trigger condition], you **MUST** produce:

[Complete template showing exactly what to generate]

Benefits:

• ✅ Copilot sees expected output format first
• ✅ Clear template to follow
• ✅ No ambiguity about deliverables

2. Added "🎯 Your Task" Section

Location: After expected output, before detailed guidance

Format:

## 🎯 Your Task

When [condition], you will:

**STEP 1**: [Action with reference to output]
**STEP 2**: [Action]
**STEP 3**: [Action]
**STEP 4**: Generate complete [deliverable] (see Expected Output above)

**Always deliver [output type]!** Do not skip or summarize.