# Python quelltest GitHub Actions workflows

> Two ways to run the Python quelltest package on pull requests: the generated quell install --pr workflow with a score comment, or a plain quell ci step.

Canonical page: https://quelltest.com/docs/guides/github-actions

For Python projects you can run quelltest on every pull request either with the workflow that `quell install --pr` generates (inline annotations plus a score comment) or with a plain `quell ci` step that only gates on a threshold. This page shows both. The npm package has its own workflow: see the [npm CI guide](https://quelltest.com/docs/npm/ci.md).

## How do I get the generated workflow?

```bash
quell install --pr
```

It writes `.github/workflows/quell.yml` (skipped if it exists). The workflow triggers on pull requests that change `**.py` files, needs `contents: read` and `pull-requests: write`, installs quelltest with `uv`, runs `quell find . --format github` so untested guard clauses show up as inline annotations, and posts or updates one pull-request comment marked `` with the Production Readiness Score, written tests and total gaps. The command reminds you to add a `QUELL_API_KEY` repository secret and to commit the file.

## How do I write a minimal gate instead?

```yaml
name: Quell
on:
  push:
    branches: [main, master]
  pull_request:

jobs:
  edge-cases:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: actions/setup-python@v5
        with:
          python-version: "3.11"
          cache: pip
      - run: pip install quelltest
      - run: quell ci src/ --threshold 0.75
```

`quell ci` exits 1 when the score is below the threshold (a fraction from 0.0 to 1.0). See [quell ci](https://quelltest.com/docs/cli/ci.md).

## How do I get inline annotations without the comment?

```yaml
      - run: pip install quelltest
      - run: quell find src/ --format github
```

Each untested guard clause becomes a workflow warning with the file, line and guard text; `quell find` exits 0, so it annotates without failing the job. Put `quell ci` in a second step to gate.

## How do I keep the report?

`quell find` writes `quell-report.json` in the working directory. Upload it as an artifact:

```yaml
      - uses: actions/upload-artifact@v4
        if: always()
        with:
          name: quell-report
          path: quell-report.json
```

## What about LLM keys?

Rule-based scans need none. If you opt in to the LLM fallback in CI, pass the provider key as a secret, for example `GROQ_API_KEY: ${{ secrets.GROQ_API_KEY }}`, and see [LLM providers](https://quelltest.com/docs/configuration/llm-providers.md) for what is sent.
