name: Eval Gate

on:
  pull_request:
  push:
    branches:
      - main

permissions:
  contents: read

jobs:
  evaluate:
    runs-on: ubuntu-latest
    steps:
      - name: Check out source
        uses: actions/checkout@v4
      - name: Run offline quality gate
        run: python3 evals/run_evals.py --threshold 0.95