Requirements for compressed air need to be analyzed for hazards. As an example, if compressed air is merely used to power or actuate equipment with no risk of contamination of food, food-contact equipment or primary packaging then the testing or filtering requirement may be minimal. If on the other hand the compressed air is used to blow, dry or move food, food-contact machinery or food-contact packaging then the hazard risk will be high. You should consider the requirements of Codex, FDA, USDA, FSMA, HACCP, HARPC, 3rd Party audits for GMP or GFSI benchmarks. I know that under the jurisdictions I have worked 1/10th micron bacterial filters had to be used, documented, dated, verified and validated for compressed air used to blow open food-contact vacuum seal bags.
Compressed air is commonly adulterated with metal oxidations, oil, water and the untold microbes that may live in that water.
Dear xylough,
i agree with yr logic. Nonetheless, based on the existing SQF threads here, a less sophisticated procedure seems to typically prevail. Maybe due to the confusing info. available from SQF.
The most rigorous, SQF recommended, general purpose, choice is installing 0.01 micron filtering capability (aka sterile filter). The least rigorous, also SQF recommended, is 0.1 micron. Why one might be preferred over the other is, afaik, unexplained. Some, but not all, manufacturers of filtration units support the former option, eg as "Best Practice". It does seem to offer the simplest (audit) route.
Alternatively, one could offer a RA. So far no SQF posts seem to have done this or, perhaps, those which do prefer to remain under the radar. Generic RAs do exist in the literature which correlate the type of process/CA usage to necessary CA quality/appropriate filtration systems, eg BCAS. These have been discussed elsewhere on this forum in some detail.
I have often wondered if SQF auditors will accept a zero filtration option at the POA [Point of application]. So far, no posted data/comments to allow assessment. 
Rgds / Charles.C