Skip to content

Getting started

Install

pip install neosloc

pipx install neosloc
# or
uv tool install neosloc

pip install git+https://github.com/neosloc/neosloc

This installs the neosloc command. python -m neosloc also works.

The static analysis needs nothing else and runs on Python 3.8 or later. LLM evaluators have their own requirements:

You want to use Install Set
Anthropic models (anthropic:…) pip install 'neosloc[agentic]' (Python ≥ 3.10) ANTHROPIC_API_KEY
OpenRouter models (openrouter:…) nothing extra OPENROUTER_API_KEY

First run

cd your/project
neosloc .

The report has four parts:

  1. Inventory: files, source files, size in tokens, languages.
  2. Dimensions: one line per dimension with its level (■■□□ 2 partial) and a one-line reason. Under it, + lines are evidence (with the file that shows it) and - lines are gaps, phrased as actions. Use -v to see all evidence.
  3. Retrofit effort: the integrability index (mean level), the person-days to bring every dimension to level 3, and a wrappability verdict.
  4. Value: classic COCOMO next to neoCOCOMO, and the share of knowledge at risk.

Use --json for a machine-readable report (schema), for example to track a project over time in CI:

neosloc --json . > neosloc.json

Assessing several projects

neosloc ~/src/service-a ~/src/legacy-b ~/src/lib-c

Reports are printed one after another (or as a JSON list with --json). Comparing old and new projects side by side is one of the main uses: a legacy application with no API but a clean CLI, tests and a readable schema often turns out cheap to wrap.

Next steps

  • Read what each dimension checks, and why.
  • Try an LLM evaluator: neosloc --review . asks a model to audit the detector's levels; neosloc --agentic . has a model attempt real integration tasks.