cd /news/developer-tools/equiv-check-that-an-ai-refactor-did-… · home topics developer-tools article
[ARTICLE · art-26131] src=github.com pub= topic=developer-tools verified=true sentiment=· neutral

Equiv, check that an AI refactor did not change what your code does

Equiv, a new open-source tool, checks that an AI refactor did not change what code does by running changed functions against previous versions on deterministically generated inputs and reporting behavioral differences. It provides signed receipts for verification, addressing the need for deterministic checks in AI-written code review. The tool supports int, str, and list[int] inputs and integrates with GitHub Actions.

read3 min publishedJun 13, 2026

An LLM should not be the only thing reviewing LLM-written code.

equiv

runs a changed function against its previous version on the same deterministically generated inputs and reports whether the behaviour changed. If it did, you get the exact input where they differ. Either way you get a reproducible, signed receipt: re-run the check on any machine and you get the same answer, byte for byte, without trusting any model's opinion.

Most code is now written by AI and reviewed by AI. A model saying "this looks fine" is not verification. A deterministic check you can re-run yourself is.

List the functions whose behaviour must be preserved across a PR in a manifest at the repository root. The format of each line is <file> : <function> : <arg types>

, where arg types are int

, str

, or list[int]

, comma separated:

src/math.py : total : int

Add the workflow at .github/workflows/equiv-review.yml

:

on: pull_request
permissions: { contents: read, pull-requests: write, id-token: write }
jobs:
  review:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
        with: { fetch-depth: 0 }
      - uses: Neelagiri65/equiv@v0.1.0
        with: { keyless: "true" }

Pin to a released tag (@v0.1.0

) rather than @main

so runs are reproducible and do not change under you.

Each PR receives a comment. Every changed function is tested against its version on the base branch. A change that preserves behaviour passes. A change that does not is reported with the input that distinguishes the two versions. That fails the check. Receipts are signed with Sigstore keyless signing, which stores no key. They can be verified with cosign

.

curl --proto '=https' --tlsv1.2 -LsSf \
  https://github.com/Neelagiri65/equiv/releases/latest/download/equiv-cli-installer.sh | sh
equiv review candidate.py reference.py <function> <arg types>
equiv verify-receipt <signed-receipt-hex>

Exit codes: 0

equivalent, 1

diverges with a printed counterexample, 2

could not check.

equiv

checks behavioural equivalence of a function against a reference, on deterministically generated inputs. This is bounded random testing, not exhaustive verification: a pass means no divergence was found on the generated inputs. It can still miss an edge case that only shows up for an input that was not generated. It does not check intent, architecture, security. It cannot judge new functionality that has no reference to compare against. A passing result means behaviour was preserved on the tested inputs. It does not mean the change is correct. Supported input types in this version are int

, str

and list[int]

.

Input generation and the verdict are computed in Rust from a fixed seed. The language runtime is used only as an evaluator and never decides anything that reaches the receipt. Receipts are identical across hosts. Receipts can be signed with a local ed25519 key or with keyless Sigstore (OIDC). The keyless path binds the signature to a verifiable CI identity rather than a stored secret. The tool is a single static binary with no runtime dependencies, prebuilt for macOS, Linux and Windows.

docs/signing-model.md

: receipt signing with ed25519 and keyless Sigstore.docs/RELEASING.md

: building prebuilt binaries with cargo-dist.crates/

: the Rust workspace (equiv-core

,equiv-engine

,equiv-review

,equiv-cli

).

License: Apache-2.0.

── more in #developer-tools 4 stories · sorted by recency
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/equiv-check-that-an-…] indexed:0 read:3min 2026-06-13 ·