# Multi-Agent Testing Prompts for MD-to-PDF Converter

Copy-paste these prompts into separate Claude instances to run parallel testing.

---

## AGENT 1: Header & TOC Edge Cases

```
I'm Agent 1 in a multi-agent testing effort for the MD-to-PDF Converter.

MY WORKSPACE: /Users/williammarceaujr./dev-sandbox/projects/md-to-pdf/testing/agent1/
CRITICAL: I will ONLY work in my workspace. I will NOT touch other agents' workspaces.

MY FOCUS: Table of contents generation and header handling

TEST SCENARIOS:
1. Duplicate headers - Multiple headers with identical text
2. Special characters in headers - Headers containing ?, !, @, #, $, %, &, *, (, ), [, ], {, }
3. Very long headers - Headers exceeding 100 characters
4. Deep nesting - All 6 header levels (H1-H6) with proper hierarchy

MY TASKS:
1. Create markdown test files in my workspace for each scenario
2. Run the MD-to-PDF converter on each test file using the wrapper script
3. Review generated PDFs for issues
4. Document findings in /Users/williammarceaujr./dev-sandbox/projects/md-to-pdf/testing/agent1/FINDINGS.md

SEVERITY CATEGORIES:
- CRITICAL: Crashes, data loss, broken core functionality
- IMPORTANT: Incorrect rendering, missing features
- NICE-TO-HAVE: Minor improvements, polish

EXPECTED BEHAVIOR:
- Unique anchor IDs for duplicate headers
- Sanitized anchors for special characters
- Properly wrapped long headers
- Complete TOC hierarchy with correct indentation

HOW TO RUN CONVERTER:
Use the wrapper script (handles library paths automatically):
/Users/williammarceaujr./dev-sandbox/projects/md-to-pdf/src/convert.sh input.md output.pdf

Example:
cd /Users/williammarceaujr./dev-sandbox/projects/md-to-pdf/testing/agent1/
/Users/williammarceaujr./dev-sandbox/projects/md-to-pdf/src/convert.sh test1.md test1.pdf
```

---

## AGENT 2: Content Formatting Edge Cases

```
I'm Agent 2 in a multi-agent testing effort for the MD-to-PDF Converter.

MY WORKSPACE: /Users/williammarceaujr./dev-sandbox/projects/md-to-pdf/testing/agent2/
CRITICAL: I will ONLY work in my workspace. I will NOT touch other agents' workspaces.

MY FOCUS: Markdown element rendering

TEST SCENARIOS:
1. Complex tables - Multi-column tables with merged cells, alignment variations
2. Nested lists - Deeply nested ordered/unordered lists (4+ levels)
3. Code blocks - Multiple languages (Python, JavaScript, Bash, SQL), very long lines
4. Mixed content - Blockquotes containing code, lists, and tables

MY TASKS:
1. Create markdown test files in my workspace for each scenario
2. Run the MD-to-PDF converter on each test file using the wrapper script
3. Review generated PDFs for issues
4. Document findings in /Users/williammarceaujr./dev-sandbox/projects/md-to-pdf/testing/agent2/FINDINGS.md

SEVERITY CATEGORIES:
- CRITICAL: Crashes, data loss, broken core functionality
- IMPORTANT: Incorrect rendering, missing features
- NICE-TO-HAVE: Minor improvements, polish

EXPECTED BEHAVIOR:
- Tables render without overflow issues
- List indentation maintains clarity
- Code blocks preserve formatting and syntax highlighting
- Mixed content renders cleanly

HOW TO RUN CONVERTER:
Use the wrapper script (handles library paths automatically):
/Users/williammarceaujr./dev-sandbox/projects/md-to-pdf/src/convert.sh input.md output.pdf

Example:
cd /Users/williammarceaujr./dev-sandbox/projects/md-to-pdf/testing/agent2/
/Users/williammarceaujr./dev-sandbox/projects/md-to-pdf/src/convert.sh test.md test.pdf
```

---

## AGENT 3: Image & Link Handling

```
I'm Agent 3 in a multi-agent testing effort for the MD-to-PDF Converter.

MY WORKSPACE: /Users/williammarceaujr./dev-sandbox/projects/md-to-pdf/testing/agent3/
CRITICAL: I will ONLY work in my workspace. I will NOT touch other agents' workspaces.

MY FOCUS: External resources and hyperlinks

TEST SCENARIOS:
1. Missing images - References to non-existent image files
2. Various image formats - PNG, JPG, GIF, SVG
3. External links - HTTP/HTTPS URLs
4. Internal anchors - Links to headers within document

MY TASKS:
1. Create markdown test files in my workspace for each scenario
2. Run the MD-to-PDF converter on each test file using the wrapper script
3. Review generated PDFs for issues
4. Document findings in /Users/williammarceaujr./dev-sandbox/projects/md-to-pdf/testing/agent3/FINDINGS.md

SEVERITY CATEGORIES:
- CRITICAL: Crashes, data loss, broken core functionality
- IMPORTANT: Incorrect rendering, missing features
- NICE-TO-HAVE: Minor improvements, polish

EXPECTED BEHAVIOR:
- Graceful handling of missing images (warning, not crash)
- All image formats render correctly
- External links are clickable
- Internal anchor links navigate correctly

HOW TO RUN CONVERTER:
Use the wrapper script (handles library paths automatically):
/Users/williammarceaujr./dev-sandbox/projects/md-to-pdf/src/convert.sh input.md output.pdf

Example:
cd /Users/williammarceaujr./dev-sandbox/projects/md-to-pdf/testing/agent3/
/Users/williammarceaujr./dev-sandbox/projects/md-to-pdf/src/convert.sh test.md test.pdf
```

---

## AGENT 4: Large Documents & Performance

```
I'm Agent 4 in a multi-agent testing effort for the MD-to-PDF Converter.

MY WORKSPACE: /Users/williammarceaujr./dev-sandbox/projects/md-to-pdf/testing/agent4/
CRITICAL: I will ONLY work in my workspace. I will NOT touch other agents' workspaces.

MY FOCUS: Scalability and edge case file sizes

TEST SCENARIOS:
1. Very large document - 100+ headers, 500+ lines
2. Empty document - No headers, minimal content
3. No headers - Document with content but no headers (TOC should be empty)
4. UTF-8 special characters - Emojis, accented characters, CJK characters

MY TASKS:
1. Create markdown test files in my workspace for each scenario
2. Run the MD-to-PDF converter on each test file using the wrapper script
3. Review generated PDFs for issues
4. Document findings in /Users/williammarceaujr./dev-sandbox/projects/md-to-pdf/testing/agent4/FINDINGS.md

SEVERITY CATEGORIES:
- CRITICAL: Crashes, data loss, broken core functionality
- IMPORTANT: Incorrect rendering, missing features
- NICE-TO-HAVE: Minor improvements, polish

EXPECTED BEHAVIOR:
- Large documents generate without timeout/memory errors
- Empty documents handle gracefully
- No-header documents skip TOC generation
- UTF-8 characters render correctly

HOW TO RUN CONVERTER:
Use the wrapper script (handles library paths automatically):
/Users/williammarceaujr./dev-sandbox/projects/md-to-pdf/src/convert.sh input.md output.pdf

Example:
cd /Users/williammarceaujr./dev-sandbox/projects/md-to-pdf/testing/agent4/
/Users/williammarceaujr./dev-sandbox/projects/md-to-pdf/src/convert.sh test.md test.pdf
```

---

## Instructions for William

1. Open 4 separate Claude instances (4 browser tabs or windows)
2. Copy-paste each agent prompt into a separate instance
3. Let all 4 agents work in parallel
4. Wait for all agents to complete (60-90 minutes)
5. Return to this session for consolidation

After all agents complete, I will:
1. Review all 4 FINDINGS.md files
2. Create consolidated-results/CONSOLIDATED-FINDINGS.md
3. Categorize issues by severity
4. Prioritize fixes
5. Implement critical fixes
6. Update directive and documentation
