quelltest 0.3.0

Verified AI test synthesis — proves every generated test catches real bugs

quelltest

Auto-generate verified killing tests for survived mutants from mutmut and Stryker.

Quell reads your mutation testing results, generates pytest assertions that kill each surviving mutant, verifies them, and injects them directly into your test files — without touching your formatting.

How it works

1. Scan — reads .mutmut-cache or Stryker JSON. Finds all survived mutants.
2. Generate — deterministic rule-based generators handle 9 operator types. No LLM call needed for the common cases.
3. Verify — every generated test runs against the live mutant in isolation. Tests that don't kill are discarded.
4. Inject — verified tests are written using libcst, a lossless concrete syntax tree parser. Comments, spacing, and formatting are preserved exactly. Source files are backed up first.

Installation

pip install quelltest

Requires Python 3.11+.

Quick start

# Run your mutation tool first
mutmut run
# or
npx stryker run --reporters=json

# Scan survived mutants
quell scan --source mutmut

# Generate and inject killing tests
quell fix

# Preview without writing
quell fix --dry-run

# Auto-fix all without prompts
quell auto

Supported mutation operators

Nine operators have deterministic rule-based generators — no network call required:

Operator	Example
BOUNDARY_SHIFT	> → >=
ARITHMETIC_OP	+ → -
LOGICAL_OP	and → or
COMPARISON_OP	== → !=
RETURN_VALUE	return x → return None
STATEMENT_DEL	statement removed
CONSTANT_MUTATION	0 → 1
DECORATOR_REMOVAL	decorator stripped
COLLECTION_OP	append → remove

UNKNOWN operators fall back to an LLM if a provider is configured.

Configuration

quell init   # adds [tool.quell] to pyproject.toml

[tool.quell]
llm_provider = "anthropic"         # "anthropic" | "openai" | "ollama"
llm_model    = "claude-sonnet-4-6"
max_verification_attempts   = 3
verification_timeout_seconds = 30
auto_write = false

Set your LLM API key (only needed for UNKNOWN operators):

export ANTHROPIC_API_KEY=sk-ant-...
# or
export OPENAI_API_KEY=sk-...

For a fully local/offline setup, use Ollama:

quell fix --llm ollama   # requires ollama running locally

Privacy

• Your code is never sent to any server unless you configure an LLM provider.
• LLM is called only for UNKNOWN operators — the rule engine handles everything else.
• All source file mutations are done locally in a subprocess and reverted afterwards.

Adapters

Tool	Format	Status
mutmut	.mutmut-cache (SQLite)	✅ Supported
Stryker (JS/TS)	reports/mutation/mutation.json	✅ Supported
PIT (Java)	XML	🔜 Planned

Project structure

quell/
├── cli.py              # Typer CLI entry point
├── adapters/           # mutmut + Stryker result parsers
├── core/
│   ├── analyzer.py     # classifies mutation operators from AST diffs
│   ├── generator.py    # rule-based test generators for 9 operators
│   ├── verifier.py     # runs tests against live mutants in subprocess
│   └── writer.py       # libcst-based test file injection
├── llm/                # LLM client + Anthropic / OpenAI / Ollama providers
└── ui/                 # Rich terminal UI (progress, diffs, console)

Development

# Clone and install
git clone https://github.com/shashank7109/quelltest_lib.git
cd quelltest_lib
uv sync

# Run tests
uv run pytest tests/ -v

# Lint
uv run ruff check . --fix

Related

• quell_frontend — Next.js dashboard
• quell_backend — FastAPI backend

License

MIT — see LICENSE