Peptides in general have high molecular weights etc. Thus peptidomimetics usually violet one or two of Lipinski rule of five. I have found it useful to develop my own filter taking into consideration some aspects of the rule of five like (Molecular-weight, H-bond acceptors/donors, logP and rotatable bonds) depending on the set of compounds at hand. As a starting point the OpenEye FILTER program has a drug-like filter that one can modify according to ones needs.

Oral bioavailability of peptide molecules is difficult but not impossible. For modelling- see paper "Predictive approaches to increase absorption of compounds during lead optimisation" - cmr/logp analysis. Many peptides may requires active trasport mechanisms that are difficult to predict. Cell permeability can be driven by endocytic approaches that CPP's adopt to get into cells and can be predicted to a degree (http://crdd.osdd.net/raghava/cellppd/multi_pep.php)

If you go back to original Lipinsky work, he just set up the Ro5 for filtering-off peptide or peptide-like hits from hit-lists resulting from HTS. (Weber's rules have about the same goal). So, they are relevant to your question but not absolute as many exceptions exists.

I suppose testing your compouds for active transport and proteolysis resistance could also help.

you may find this helpful (an important resource for peptide based therapeutics and peptide drug designing)

SATPdb: A database of structurally annotated therapeutic peptides.

http://crdd.osdd.net/raghava/satpdb/

paper link

http://nar.oxfordjournals.org/content/early/2015/11/01/nar.gkv1114.full