Python reference implementation of the Elephant Visibility Index (EVI).
Drop-in scorer that takes a domain and returns the same EVI breakdown the Elephant Accountability audit engine produces. Useful for self-audits, CI gating, and contribution to the spec.
Pre-release v0.1.0 — Monday 2026-05-12. This package is a scaffold; the scorer methods raise NotImplementedError until the canonical scoring code lands. The canonical scorer lives in the Elephant Accountability orchestrator. This repo is the public, pip-installable mirror.
pip install evifrom evi import audit
result = audit("example.com")
print(result.score) # 0-100
print(result.tier) # "bronze" | "silver" | "gold" | None
print(result.axis_scores) # per-axis breakdown
print(result.surfaces) # detected surfaces with raw evidenceThe package version follows semver and pins to a specific EVI spec version:
evi 0.x→ tracksevi-spec v2.x(scoring weights and rules from the v2 draft).evi 1.x→ first stable release, locks in v2-final scoring.evi 2.x→ tracks future spec major versions.
Material spec changes that affect scored outputs bump the package major version.
pip install -e ".[test]"
pytestMIT.
Issues and PRs welcome. The spec lives at elephant-accountability/evi; this repo is the implementation. Spec changes go to that repo first; implementation follows.
Maintained by Elephant Accountability.