Quality assurance for a pull request: gate on the coverage of the lines it changed, and the findings it introduced.
Point Brimyr at any repository and it works out the rest. It detects the ecosystem, runs the test suite with coverage instrumentation on, and gates on the coverage of the diff rather than the whole codebase. No per-repo configuration, no server, no SaaS account, and no coverage report you have to produce first.
Documentation · Setup · Action reference · Marketplace
# .github/workflows/quality.yml
name: Quality
on: [pull_request]
permissions:
contents: read
pull-requests: write # only needed for pr_comment
jobs:
brimyr:
runs-on: ubuntu-latest
steps:
- uses: magmamoose/brimyr@v1 # that is the whole configuration
with:
pr_comment: 'true' # off by default- Runs your tests — detects the ecosystem and turns coverage instrumentation on itself. Every other patch-coverage tool asks you to bring your own report, and across eight repos in four languages that wiring is the project.
- Gates on the diff — the coverage of the lines this PR changed, matching diff-cover semantics deliberately. A large existing gap never blocks anyone.
- Gates on net-new findings — with
quality: 'true', classifies linter findings against that same diff via Chargate. - One consolidated PR comment — updated in place, never stacked. Opt in with
pr_comment: 'true'. - Ships the trend — optional SonarQube export, without letting it gate.
- Runs entirely on your runner — free, Apache-2.0, no backend.
Brimyr executes the pull request's own test code on the runner. That is inherent to running tests, and it is why the default workflow trigger is
pull_requestrather thanpull_request_target: the former gives a fork's code no access to your secrets. Do not change that without reading Setup.
| Input | Default | What it does |
|---|---|---|
threshold |
80 |
Patch-coverage percentage that blocks below it. |
quality |
false |
Also gate on net-new quality findings, running Chargate as a nested step. |
ecosystem |
auto-detected | Force one or more: python, javascript (jest), vitest, node-test, dotnet, java, go, shell. |
mode |
auto |
auto from the event · pr (diff gate) · baseline. |
pr_comment |
false |
Post one consolidated PR comment. Needs pull-requests: write. |
sonar_url |
— | Set to export the trend to SonarQube. |
All inputs and outputs → Action reference
| Setup and usage | Workflows, the pre-push hook, thresholds |
| Patch coverage · Quality findings | What counts as covered, and what counts as net-new |
| .NET · Java / JVM · Go · Shell / bats | Per-ecosystem notes |
| Action reference · CLI reference | Every input, output and command |
| Architecture · Troubleshooting | How it works, and what to do when it doesn't |
Brimyr gates tests and coverage · Chargate gates security and lint · Diatreme releases what passes · Tremvok deploys and verifies
Brimyr and Chargate answer different questions about the same diff. Chargate classifies the findings; Brimyr decides whether the PR is adequately tested, and can fold Chargate's verdict into its own.
Pin @v1 for the floating major, or a tag or SHA to freeze.
Report a vulnerability · Contributing · Apache-2.0, see LICENSE.