Getting Started¶

Installation¶

pip install bagx

Optional extras¶

pip install bagx[mcap]   # .mcap file support
pip install bagx[llm]    # LLM-powered ask command (Anthropic / OpenAI)

From source¶

git clone https://github.com/rsasaki0109/bagx.git
cd bagx
pip install -e ".[dev]"

ROS2 dependency¶

bagx works with or without ROS2:

With ROS2: Full message deserialization via rosbag2_py
Without ROS2: .db3 files are read via SQLite with built-in CDR parsers for common types (NavSatFix, Imu, PoseStamped, Odometry, PointCloud2, TFMessage)

First use¶

# Show bag info
bagx info your_bag.db3

# Evaluate quality
bagx eval your_bag.db3

# Evaluate with a built-in custom-rule plugin for your stack
bagx rules list
bagx eval your_bag.db3 --rules warehouse_bot

# Export to JSON for inspection
bagx eval your_bag.db3 --json report.json

# Run a benchmark suite
bagx benchmark benchmarks/open_data_suite.json
bagx benchmark benchmarks/non_slam_suite.json

CLI options¶

bagx --version          # Show version
bagx --verbose eval ... # Debug logging
bagx --quiet eval ...   # Errors only
bagx --help             # Show all commands

Supported formats¶

Format	Extension	Requirement
SQLite	`.db3`	None (built-in)
MCAP	`.mcap`	`pip install bagx[mcap]`
Directory	folder with `.db3`/`.mcap`	Auto-detected
Multi-segment	folder with multiple files	Aggregated automatically