VeriBiota Integration Guide¶

This shows how to call VeriBiota from CI or Python with stdin piping and stable exit codes.

CLI expectations¶

0 → profile holds (status: "passed").
2 → profile checked but failed (status: "failed").
1 → malformed input or internal error.

Profiles:

global_affine_v1
edit_script_v1
edit_script_normal_form_v1
prime_edit_plan_v1
pair_hmm_bridge_v1
vcf_normalization_v1

Each verdict JSON includes profile_version, theorems, and an engine block (veribiota_version, lean_version, build_id).

GitHub Actions snippet (download binary)¶

- name: Download VeriBiota
  run: |
    VERIBIOTA_TAG=v0.2.1   # TODO: pin to the release you adopt
    PLATFORM=linux-amd64
    BUNDLE="veribiota-${VERIBIOTA_TAG}-${PLATFORM}"
    curl -L "https://github.com/OmnisGenomics/VeriBiota/releases/download/${VERIBIOTA_TAG}/${BUNDLE}.tar.gz" -o "${BUNDLE}.tar.gz"
    tar -xzf "${BUNDLE}.tar.gz"
    chmod +x "${BUNDLE}/veribiota"
    echo "VERIBIOTA_EXE=$PWD/${BUNDLE}/veribiota" >> $GITHUB_ENV
    # Optional (only needed if you move the binary away from the bundle directory):
    echo "VERIBIOTA_DATA_DIR=$PWD/${BUNDLE}" >> $GITHUB_ENV

- name: Run profile check
  run: |
    cat examples/profiles/global_affine_v1/match.json | "$VERIBIOTA_EXE" check alignment global_affine_v1 -

Python helper (stdin piping)¶

Install/ship veribiota_py (in this repo):

from veribiota_py import check_alignment_global_affine_v1, check_edit_script_v1

instance = {
  "seqA": "A",
  "seqB": "A",
  "scoring": {"match": 2, "mismatch": -1, "gap_open": -2, "gap_extend": -1},
  "witness": {"score": 2, "trace": [{"op": "M"}]},
}

verdict = check_alignment_global_affine_v1(instance)
assert verdict["exit_code"] == 0
assert verdict["status"] == "passed"

Control which executable is used via VERIBIOTA_EXE (default: veribiota on PATH). Inputs can always be piped via - to avoid temp files.