version: "1.0.0" name: extract-behavior description: Extracts the intended behavior of a Python module or package into BEHAVIOR.md as reviewable tables. Use before writing tests. argument-hint: <path/to/module.py | path/to/package/> disable-model-invocation: true

Extract Behavior

Extract the intended behavior of $ARGUMENTS into BEHAVIOR.md.

Steps

1. Collect the target modules:

• If $ARGUMENTS is a .py file, the target is that module.
• If it's a directory, the targets are all .py files under it,

recursively, in path order. Skip test files and files whose names start with an underscore (_).

1. Carefully read the package that contains the targets and map it: list

each module's public callables, their parameters, branches, error paths, and external dependencies.

1. For each target module, write its section group in BEHAVIOR.md at

the project root. Start the group with a ` # <module path> ` heading, then write one section per public callable. Include every decision point, boundary, error, and edge case. Replace any existing group for the module, and leave the other groups alone.

1. Report the modules and sections you wrote, and every row marked

NEEDS REVIEW.

Don't write tests or edit any code.

Template

Use this template for every section:

## <callable>(<parameters>)

<One sentence on what the callable returns or does.>

Source: <docstring, comment, or requirement>.

Rules

• Add one column per parameter. For stateful code, add a given column

first to record the starting state.

• One row per case. Write boundaries as concrete rows on both sides,

never as ranges.

• Start each note with happy, boundary, edge, or error.
• Mark values inferred only from the implementation as NEEDS REVIEW.