What is the purpose of the `rake cover_me:report` task?

It's not really clear to me what this does. Is it that it must be run after your tests in order for an accurate report to be generated? I find it odd that after running my test suite, my coverage report is opened in my browser (89% coverage, thankyouverymuch), but if I run rake cover_me:report after that, the coverage report is opened in my browser again, but with 92% coverage.

This would make sense to me if every file had increased somewhat in coverage (as my test suite runs rspec tests and then cucumber tests, and the initial coverage report is opened as the rspec tests are completed), but the numbers just seem to change after running rake cover_me:report: some go up, some go down. Doesn't that seem somewhat odd to you?

Also, a related question: is it possible to programmatically determine the overall coverage percentage? I would love to fail my build if it dips below some arbitrary threshold.

Start with the rake cover_me:report task and trace how it combines coverage from the RSpec and Cucumber runs. Compare the reports produced before and after that task, then document the task's purpose, why percentages differ, and whether the overall percentage can be retrieved for a build check.