Embed the lifecycle gate in your CI/CD

Stop discovering a model retirement from a 4xx in production. LLMIntel turns the lifecycle feed into a build gate: scan your repo for referenced model ids and fail the pipeline before a retired model ships.

The basic gate and the API are free. Want it truly hands-off? $5/mo runs llmintel sync in CI to auto-watch the models your code actually uses, then pushes webhook, Slack, and PagerDuty alerts the moment one is deprecated or retired — no watchlist to curate. Wire it once, then forget it. See plans.

1
Mint an API key
Create an account-scoped key from your dashboard. The free tier is enough to gate one repo.
2
Drop the gate into your pipeline
Add the step below as LLMINTEL_API_KEY secret. It scans your code for model ids and queries lifecycle state.
3
Builds fail before models do
The check exits non-zero on any retired or past-due model — so you migrate on your schedule, not the provider’s shutoff date.

.github/workflows/model-lifecycle.yml

# .github/workflows/model-lifecycle.yml
name: Model lifecycle gate
on: [pull_request]

jobs:
  lifecycle-gate:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v6
      - uses: hivemindunit/llmintel.ai/packages/cli@main
        with:
          api-key: ${{ secrets.LLMINTEL_API_KEY }}
          paths: "src config"     # scanned for model ids/aliases
          warn-days: "90"          # warn when retiring within 90 days
          fail-on-warn: "false"    # set true to block on deprecations too