I tried PDF::API2, CAM::PDF, and PDF::Reuse to implement an option for my existing PDF writer software to solve the task of embedding other PDFs into the file that is produced.
PDF::API2 and PDF::Reuse have problems with newer PDF versions, as stated above in this thread.
My results with CAM::PDF are good; now I am looking for a modification of this module to embed pages from other documents into pages of the generated document instead of replacing them.
Besides, PDF::Reuse should be modified to be object oriented, or at least to not export all its functions by default. It relies on internal global variables what should by taken into account when processing several files. It's well written, though.
I as well consider using ghostscript; maybe it helps downgrading newer PDFs as it helps breaking document protection - another problem you might encounter trying to append external PDFs to files your scripts create.
The PDF writer software I use is authored by myself and pretty lengthy as I basically implemented a browser in Perl, reading HTML templates, creating SQL queries from these, downloading data and images from the database (using HTTP/mod_plsql) and doing all the box layout calculation.
PDF generation is finally done with PDFlib.
(Nowadays, I might favor using XSL:FO instead of doing all the calculation myself - too long a way to go.)
So post processing of PDF files maybe challenging, or require commercial software to handle them if you can't allow for PDF version limitations.