I developed part of a method developed for isolating bacterial DNA from human serum samples, but then we shifted what we were doing, so I never finished. I had some success using the QIAgen DNA Mini Kit and adding a bead beating step, but I was spiking the serum with E coli, which is not difficult to lyse. You may have more difficulty if you are targeting a different species. I don't know anything about csf, but the kit might be a place to start.

Standard Alkaline lysis method is best suitable for extraction of bacterial DNA. Its gives good yield as well as quality of DNA to be used as template for its amplification using conventional PCR.