wehub-resource-sync
120 lines
2.7 KiB
Markdown
120 lines
2.7 KiB
Markdown
# MLflow Claude Code Integration
|
|
|
|
This module provides automatic tracing integration between Claude Code and MLflow.
|
|
|
|
## Module Structure
|
|
|
|
- **`config.py`** - Configuration management (settings files, environment variables)
|
|
- **`hooks.py`** - Claude Code hook setup and management
|
|
- **`cli.py`** - MLflow CLI commands (`mlflow autolog claude`)
|
|
- **`tracing.py`** - Core tracing logic and processors
|
|
- **`hooks/`** - Hook implementation handlers
|
|
|
|
## Installation
|
|
|
|
```bash
|
|
pip install mlflow
|
|
```
|
|
|
|
## Usage
|
|
|
|
Set up Claude Code tracing in any project directory:
|
|
|
|
```bash
|
|
# Set up tracing in current directory
|
|
mlflow autolog claude
|
|
|
|
# Set up tracing in specific directory
|
|
mlflow autolog claude -d ~/my-project
|
|
|
|
# Set up with custom tracking URI
|
|
mlflow autolog claude -u file://./custom-mlruns
|
|
mlflow autolog claude -u sqlite:///mlflow.db
|
|
|
|
# Set up with Databricks
|
|
mlflow autolog claude -u databricks -e 123456789
|
|
|
|
# Check status
|
|
mlflow autolog claude --status
|
|
|
|
# Disable tracing
|
|
mlflow autolog claude --disable
|
|
```
|
|
|
|
## How it Works
|
|
|
|
1. **Setup**: The `mlflow autolog claude` command configures Claude Code hooks in a `.claude/settings.json` file
|
|
2. **Automatic Tracing**: When you use the `claude` command in the configured directory, your conversations are automatically traced to MLflow
|
|
3. **View Traces**: Use `mlflow server` to view your conversation traces
|
|
|
|
## Configuration
|
|
|
|
The setup creates two types of configuration:
|
|
|
|
### Claude Code Hooks
|
|
|
|
- **PostToolUse**: Captures tool usage during conversations
|
|
- **Stop**: Processes complete conversations into MLflow traces
|
|
|
|
### Environment Variables
|
|
|
|
- `MLFLOW_CLAUDE_TRACING_ENABLED=true`: Enables tracing
|
|
- `MLFLOW_TRACKING_URI`: Where to store traces (defaults to local `.claude/mlflow/runs`)
|
|
- `MLFLOW_EXPERIMENT_ID` or `MLFLOW_EXPERIMENT_NAME`: Which experiment to use
|
|
|
|
## Examples
|
|
|
|
### Basic Local Setup
|
|
|
|
```bash
|
|
mlflow autolog claude
|
|
cd .
|
|
claude "help me write a function"
|
|
mlflow server --backend-store-uri sqlite:///mlflow.db
|
|
```
|
|
|
|
### Databricks Integration
|
|
|
|
```bash
|
|
mlflow autolog claude -u databricks -e 123456789
|
|
claude "analyze this data"
|
|
# View traces in Databricks
|
|
```
|
|
|
|
### Custom Project Setup
|
|
|
|
```bash
|
|
mlflow autolog claude -d ~/my-ai-project -u sqlite:///mlflow.db -n "My AI Project"
|
|
cd ~/my-ai-project
|
|
claude "refactor this code"
|
|
mlflow server --backend-store-uri sqlite:///mlflow.db
|
|
```
|
|
|
|
## Troubleshooting
|
|
|
|
### Check Status
|
|
|
|
```bash
|
|
mlflow autolog claude --status
|
|
```
|
|
|
|
### Disable Tracing
|
|
|
|
```bash
|
|
mlflow autolog claude --disable
|
|
```
|
|
|
|
### View Raw Configuration
|
|
|
|
The configuration is stored in `.claude/settings.json`:
|
|
|
|
```bash
|
|
cat .claude/settings.json
|
|
```
|
|
|
|
## Requirements
|
|
|
|
- Python 3.10+ (required by MLflow)
|
|
- MLflow installed (`pip install mlflow`)
|
|
- Claude Code CLI installed
|