quell reproduce

Takes a plain-English bug description and uses an LLM to generate a pytest test that reproduces it. The test is verified to currently fail (confirming the bug exists) before being written to disk.

Usage

quell reproduce DESCRIPTION [OPTIONS]

Arguments

Argument	Description
`DESCRIPTION`	Bug description in plain English

Options

Option	Default	Description
`--file`	`None`	Target source file to focus on
`--root`	`.`	Project root directory

Examples

# Reproduce a bug from a description
quell reproduce "payment accepts zero amount silently"

# Focus on a specific file
quell reproduce "discount applies when price equals threshold" --file src/calculator.py

How it works

The description is sent to your configured LLM provider along with relevant source context.
The LLM generates a pytest test that exercises the described behavior.
Quell verifies the test currently fails (confirming the bug is present).
The failing test is written to your test file.

The test will pass once the bug is fixed — giving you a clear signal that your fix is complete.

Requires LLM

quell reproduce always uses an LLM because bug descriptions are unstructured natural language. Configure your provider in pyproject.toml:

[tool.quell]
llm_provider = "anthropic"   # "anthropic" | "openai" | "ollama"
llm_model = "claude-sonnet-4-6"

And set your key:

export ANTHROPIC_API_KEY=sk-ant-...

Exit codes

Code	Meaning
`0`	Test generated and written
`1`	Could not generate a verified reproduction test