Avante - Example

1.1 File Structure

.
├── prompts/
│   └── npc_fix.avante.md
├── new_script/
│   ├── __init__.py
│   ├── npc_extractor.py
│   └── json_generator.py
└── tests/
    └── test_npc_extractor.py

1.2 Creating the Prompt Template

# File: prompts/npc_fix.avante.md
 
```avante
## NPC Parsing Fix Requirements
 
### Current Issues
1. Failed name extraction in test_npc_extractor_1
2. Partial trait capture (misses multi-line entries)
3. Incorrect skill value parsing
4. ST/DX block detection failures
 
### Code Context
{{FILE:new_script/npc_extractor.py}}
{{FILE:tests/test_npc_extractor.py}}
 
### Test Data
```python
# Sample failing test case
assert parsed_npc['name'] == 'Swamp Lion'  # Actual: ''

Success Criteria

• All tests pass with original input structure
• Handle wrapped lines in traits/skills
• Support both ST:14 and ST\n14 formats

2. AI-Assisted Debugging Workflow

2.1 Interactive Session

# Start LunarVim
lvim .
 
# In normal mode:
<Space>a f  # Add file context
<Space>a p  # Create new prompt

2.2 Key avante.nvim Commands

Command	Action
:AvanteFileContext	Select project files for context
:AvanteDiff	View proposed code changes
:AvanteAsk	Submit custom query

2.3 Example AI Interaction

User Input:

Fix the regex patterns to handle multi-line traits while preserving test structure

AI Response:

# npc_extractor.py
 
- m = re.search(rf"{field}:\s*(.+)", npc_sheet_text)
+ m = re.search(rf"{field}:\s*((?:.|\n)+?)(?=\n\w+:|$)", npc_sheet_text)
 
# test_npc_extractor.py
 
-       'Traits': ['Night Vision 5', 'Quadruped', 'Sharp Claws', 'Sharp Teeth'],
+       'Traits': ['Night Vision 5', 'Quadruped', 'Sharp Claws', 
+                 'Sharp Teeth', 'Wild Animal'],

3. Key Technical Solutions

3.1 Multiline Regex Pattern

# Before (fails on wrapped lines)
r"Traits:\s*(.+)"  
 
# After (captures until next section)
r"Traits:\s*((?:.|\n)+?)(?=\n\w+:|$)"

3.2 Name Detection Improvement

# Original
r"^(?P<name>[A-Za-z\s]+)$"
 
# Enhanced with lookahead
r"^([A-Z][a-z]+(?: [A-Z][a-z]+)*)\n(?=HP:|ST)"

3.3 Skill Parsing Upgrade

# Handle multi-line skills
skills = re.split(r'[;\n]', value)
skills_dict = {}
for skill in skills:
    if '-' in skill:
        name, val = skill.split('-', 1)
        skills_dict[name.strip()] = int(val.strip())

4. Validation Workflow

4.1 Test Execution

:LvimTerminal python -m pytest tests/ -v --pdb

4.2 Debug Integration

-- Configure DAP for Python
require('dap').adapters.python = {
  type = 'executable',
  command = 'python',
  args = { '-m', 'debugpy.adapter' }
}

5. Best Practices

1. Incremental Fixes: Address one test failure at a time
2. Context Isolation: Use separate avante prompts for different components
3. Diff Verification: Always review :AvanteDiff before applying changes
4. Test Preservation: Maintain original test structure during fixes