ci.yml

name: CI

# Scope: keyless, self-contained checks only. Everything that needs API keys or
# local benchmark data — eval:* and benchmark:* scripts, OmniMemEval suites,
# summary/judge providers — stays local by design.

on:
  push:
    branches: [main]
  pull_request:
    branches: [main]

concurrency:
  group: ci-${{ github.workflow }}-${{ github.ref }}
  cancel-in-progress: true

jobs:
  static:
    name: Static and package contracts
    runs-on: ubuntu-latest
    timeout-minutes: 20
    steps:
      - uses: actions/checkout@v6
      - uses: actions/setup-node@v6
        with:
          node-version: "24"
          cache: npm
      - run: npm ci
      # Subpackage builds (verify:packages) use pnpm via the dsh/dsh-nmg
      # pnpm-lock.yaml; corepack is bundled with the Node distribution.
      - name: Enable corepack pnpm
        run: |
          corepack enable
          pnpm --version
      - name: Every test suite is run by a CI job, or acknowledged
        run: npm run ci:uncovered-tests
      - name: Shared static verification contract
        run: npm run verify:static
      - name: Dependency audit
        run: npm audit --production --audit-level=high

  onlion:
    name: Service backend (onlion)
    runs-on: ubuntu-latest
    timeout-minutes: 30
    steps:
      - uses: actions/checkout@v6
      - uses: actions/setup-node@v6
        with:
          node-version: "24"
          cache: npm
          # Two manifests: the service builds its prompt tables with a root-level
          # tool, so both dependency trees are installed.
          cache-dependency-path: |
            package-lock.json
            onlion/package-lock.json
      - run: npm ci
      - name: Install service dependencies
        run: npm ci
        working-directory: onlion
      - name: Service verification contract
        run: npm run verify:onlion

  tests:
    name: Product tests and coverage
    runs-on: ubuntu-latest
    timeout-minutes: 30
    steps:
      - uses: actions/checkout@v6
      - uses: actions/setup-node@v6
        with:
          node-version: "24"
          cache: npm
      - run: npm ci
      - name: Shared product verification contract
        run: npm run verify:product-ci

  research-tests:
    name: Research characterization (non-blocking)
    # Deliberately not aggregated by `all-checks-passed`, so a failure here never blocks a merge.
    # It used to also carry `continue-on-error`, which made the job look green whatever happened:
    # a track that cannot be told apart from "never ran" is the failure mode this repository
    # forbids. Non-blocking is a branch-protection property, so the job now reports its real
    # result (a visible red check on the pull request) while the required aggregate stays green.
    runs-on: ubuntu-latest
    timeout-minutes: 20
    steps:
      - uses: actions/checkout@v6
      - uses: actions/setup-node@v6
        with:
          node-version: "24"
          cache: npm
      - run: npm ci
      - name: Shared research verification contract
        run: npm run verify:research

  node-compat:
    name: Node 22 compatibility
    runs-on: ubuntu-latest
    timeout-minutes: 20
    steps:
      - uses: actions/checkout@v6
      - uses: actions/setup-node@v6
        with:
          node-version: "22.19.0"
          cache: npm
      - run: npm ci
      - name: Shared minimum-Node verification contract
        run: npm run verify:node-compat

  chaos:
    name: Windows chaos tests
    # These reproduce Windows-specific EPERM file pins and WAL locking. One
    # supported runtime is enough here; Node compatibility is checked above.
    # The job failed once in 153 runs, on main, so it runs on main pushes
    # instead of every pull request: 2026-09-09-chaos-tests-on-main-pushes.
    if: github.event_name == 'push'
    runs-on: windows-latest
    timeout-minutes: 30
    steps:
      - uses: actions/checkout@v6
      - uses: actions/setup-node@v6
        with:
          node-version: "24"
          cache: npm
      - run: npm ci
      - name: Chaos tests (Windows)
        env:
          NMG_DAEMON_READY_TIMEOUT_MS: "60000"
        run: npm run verify:chaos

  all-checks-passed:
    name: All checks passed
    if: ${{ always() }}
    needs: [static, tests, node-compat, chaos]
    runs-on: ubuntu-latest
    steps:
      - name: Verify required jobs
        env:
          STATIC_RESULT: ${{ needs.static.result }}
          TESTS_RESULT: ${{ needs.tests.result }}
          NODE_COMPAT_RESULT: ${{ needs.node-compat.result }}
          CHAOS_RESULT: ${{ needs.chaos.result }}
        run: |
          test "$STATIC_RESULT" = success
          test "$TESTS_RESULT" = success
          test "$NODE_COMPAT_RESULT" = success
          # chaos is skipped on pull requests and required on main pushes.
          test "$CHAOS_RESULT" = success || test "$CHAOS_RESULT" = skipped