Filters let you manipulate your documents similarly to preprocessors. Unlike preprocessors, they do not modify the text before Pandoc gets hold of it, but instead, they are plugged into the text transformation that Pandoc does. Think of them as post-processors; it is not far from the truth except that Pandoc will run after they are done.
Both preprocessors and filters have strength and weaknesses. They can do the same things to your files, but some things are easier to program in a preprocessor, while some things are easier to program in a filter.
Pandoc can read from standard input and output ...
