cli

CLI

Use this sub-skill when a user wants to build or debug a stitch command-line invocation rather than write Python code.

Typical triggers

• "stitch these images from the shell"
• "what flags should I pass to stitch?"
• "I need affine mode or verbose output"
• "how do I use feature masks from the CLI?"
• "why does stitch fail in a headless session?"

What this sub-skill owns

• The public stitch entry point.
• Positional image arguments, glob patterns, and output paths.
• --affine, --verbose, --verbose_dir, --output, --output_params.
• Feature and matcher tuning flags such as --detector, --nfeatures, --matcher_type, --range_width, --confidence_threshold, --matches_graph_dot_file, --warper_type, --finder, and crop controls.
• The optional --feature_masks and --try_use_gpu flags.
• CLI troubleshooting for missing files, mask count problems, preview windows, and output selection.

What this sub-skill does not own

• Python Stitcher object configuration; use python-api.
• Verbose file interpretation and match-graph analysis; use diagnostics.
• Install/import troubleshooting; use the root troubleshooting.

Start with these bundled files

• CLI reference for verified flags and choices.
• Workflow recipes for copyable command examples.
• Troubleshooting for common CLI failures.
• Validate CLI args when you want a safe checker that validates input globs and feature-mask counts without running a full stitch.

Fast routing hints

• Need a single command to stitch a glob of files? Read the workflow recipes.
• Need to check whether a feature-mask list is valid before a full stitch? Run the bundled validator.
• Need to interpret dropped images or sparse matches? Route to diagnostics.

Common success criteria

A good answer from this sub-skill should include:

• The exact stitch command or command template.
• The meaning of the relevant flags.
• The expected output file or verbose directory.
• A recovery path if masks, globs, or preview windows fail.