Creates a github 'status' for every build which passes only if coverage hasn't fallen compared to master.

How it Works

The basic idea is, you run github-coverage-robot as part of your build process. On master branch, github-coverage-robot will upload a small JSON file to Amazon S3 with details about your code coverage. After any build, github-coverage-robot will fetch the JSON data from S3, and generate a github status for the build which will pass if coverage is as good as or better than master, and fail if it's worse than master.

Usage:

Install with:

npm --save github-coverage-robot

First, create a github access token repo and repo_public permissions. Second, create a bucket on S3 to store coverage reports, and obtain an AWS access key ID and secret access key. On your CI server, set up the following environment variables (note that you should NOT commit these to source control in an open source project):

export GITHUB_ROBOT_ACCESS_TOKEN=[github access token]
export AWS_ACCESS_KEY_ID=[...]
export AWS_SECRET_ACCESS_KEY=[...]

Then, when running your tests on the CI server, generate your coverage numbers and run:

./node_modules/.bin/github-coverage-robot \
    --bucket [s3Bucket] \
    --project [owner/repo] \
    --coverage [percent]

where s3Bucket is the name of the bucket you created to hold reports, owner/repo is the github user and repo name for your project, and percent is a float value representing the coverage of your project. For example:

./node_modules/.bin/github-coverage-robot \
    --bucket ci-reports \
    --project benbria/node-github-coverage-robot \
    --coverage 89.34

You can optionally specify --branch and --sha, but github-coverage-robot should be smart enough to figure them out in most cases.