Skip to content

semcod/testless

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

12 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

testless

AI Cost Tracking

PyPI Version Python License AI Cost Human Time Model

  • 🤖 LLM usage: $0.6007 (8 commits)
  • 👤 Human dev: ~$300 (3.0h @ $100/h, 30min dedup)

Generated on 2026-05-26 using openrouter/qwen/qwen3-coder-next


testless is a Python package that sits as an analytical layer on top of your existing test suite. It analyses test value, coverage overlap, duplication, and missing gaps, then generates LLM-ready planfile tickets that describe exactly what to remove, refactor, or add.


Features

Command What it does
testless scan Run pytest with --cov-context=test and collect metadata
testless duplicates Detect duplicate tests (coverage + AST + fixture overlap)
testless missing Suggest smoke, e2e, contract, and TestQL tests
testless planfiles Generate YAML planfile tickets for an LLM
testless doctor Answer questions: "what to remove?", "what to add?"

Quick Start

pip install testless
# or with uv:
uv add --dev testless

# 1. Scan your project
testless scan

# 2. Find duplicate tests (threshold: 85 % similarity)
testless duplicates --min-overlap 0.85

# 3. Find missing smoke / service / e2e tests
testless missing --services src/myapp

# 4. Generate LLM planfiles
testless planfiles --out .planfiles/ --with-prompts

# 5. Ask doctor questions
testless doctor "which tests should I remove?"

Configuration

Create a .testless.yml in your project root:

packages:
  - myapp

test_dirs:
  - tests

min_duplicate_score: 0.85
planfiles_dir: .planfiles
coverage_dir: .coverage_data

pytest_args:
  - --tb=short

Planfile format

Each generated ticket is a self-contained YAML file that an LLM can act on:

version: "1"
kind: duplicate
id: "DUP-3A7F2C"
title: "Remove duplicate test: test_health_ok vs test_health_status_200"
goal: "Reduce test suite redundancy without losing unique coverage"
context:
  test_a: tests/api/test_health.py::test_health_ok
  test_b: tests/api/test_health.py::test_health_status_200
evidence:
  - description: "score=0.93, coverage_overlap=0.95"
    score: 0.93
impact:
  risk: low
  expected_gains:
    - shorter test run time
    - reduced maintenance burden
tasks:
  - description: Compare test_health_ok and test_health_status_200 manually
  - description: Remove the weaker/less descriptive test
  - description: Run suite and verify no coverage regression
acceptance_criteria:
  - No decrease in unique line coverage
  - All remaining tests pass
llm_hints:
  style: minimal change
  safe_refactor: true

Architecture

src/testless/
├── cli.py                  # Click CLI
├── config.py               # .testless.yml loader
├── models/
│   ├── planfile.py         # Pydantic Planfile model
│   ├── findings.py         # Finding models
│   └── coverage_map.py     # CoverageMap model
├── collect/
│   ├── pytest_runner.py    # Run pytest, collect metadata
│   ├── coverage_loader.py  # Parse coverage.json contexts
│   ├── fixture_index.py    # AST-based fixture indexer
│   └── endpoint_inventory.py  # HTTP endpoint/service scanner
├── analyze/
│   ├── duplicate_tests.py  # Duplicate detection
│   ├── dead_tests.py       # Dead test detection
│   ├── missing_tests.py    # Missing test suggestions
│   ├── service_risk.py     # Service risk scoring
│   └── refactor_candidates.py  # Refactor detection
├── suggest/
│   ├── smoke.py            # Smoke test templates
│   ├── e2e.py              # E2E test templates
│   ├── service_tests.py    # Service contract templates
│   └── testql.py           # TestQL stub generator
├── tickets/
│   ├── builder.py          # Build Planfile from findings
│   ├── serializer.py       # Write planfiles to disk
│   └── prompts.py          # Attach LLM prompts
└── reporters/
    ├── console.py          # Coloured terminal output
    ├── markdown.py         # Markdown report
    └── json.py             # JSON report

License

Licensed under Apache-2.0.

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Packages

 
 
 

Contributors

Languages