It's more idiomatic to pass the error object directly to c.Fatal() instead of err.Error(). The testing harness will handle formatting it as a string, and this preserves the error type information.

This function can be simplified. The comments are not very helpful, and using regexp.Match would be more concise and idiomatic Go. regexp.Match directly returns (bool, error), which matches the function signature, and handles potential regexp compilation errors gracefully unlike regexp.MustCompile which would panic.

func fileContainsPattern(path string, searchPattern string) (bool, error) {
	file, err := os.ReadFile(path)
	if err != nil {
		return false, err
	}
	return regexp.Match(searchPattern, file)
}

The variable failConfig is shadowed inside this function. The global failConfig is a *conf.UserData, while the local failConfig returned by failConfig.Render() is a *conf.Conf. This can be confusing. Renaming the local variable would improve clarity.

renderedConfig, err := failConfig.Render(conf.FailWarnings)
	if err != nil {
		return errors.Wrapf(err, "creating invalid FIPS config")
	}

	// Create a temporary log file
	consoleFile := c.H.TempFile("console-")

	// Instruct builder to use it
	builder.ConsoleFile = consoleFile.Name()
	builder.SetConfig(renderedConfig)