I wanted to do the same thing a while back and found that there were not many articles about using rocks from the wild in your fish tank. Most people just say "go spend the 5-10 per pound for slate from the LFS or homedepot". I don't have 30-75 dollars for rocks just yet so this was the process I settled on.
I collected generally dark colored rocks with little or no shiny flecks or oxidation (rust). The shiny spots are often calcium which will break down over time and change you PH and water hardness. Chances are pretty good that a light/white rock is going to have calcium in it also so I just avoid. The rust is obviously an indication of a ferrous metal which apparently can react with the amonia in the water and release an atrocious acid. I also try to avoid soft rocks that you can easily scratch deeply with another rock or even your fingernail Apparently these are bad too. (This is what I have gathered from different articles on the web).
After I collected my rocks I scrubbed them well in warm water with a brush. A wirebruch would work great I just used a toothbrush.
After they were clean I drop each rock or if I had rocks that appeared to be the same kind of stone into a plastic container with tap water. I let the rocks soak between 24 and 48 hours and then checked the PH (my tapwater is 7). If I saw any change at all I pitched the rocks or retested; I think one of the tupperware dishes I used may have housed pasta sauce prior to usage so there was some acid leach from the container, second pass it was fine.
After the test is done I take passing rocks and boil them in a stockpot for about 15-30 minutes. I try to make sure the entire rock to the core is hot.
I then cool and add to the tank.
I have added about 10-15 large rocks to my aquarium over the last month with no problems. I monitor PH weekly for now, just in case.
I am not sure if this is useful as you are after small rocks. Perhaps you could look through your rocks to figure out which ones look similar and then process in batches. If a container has a PH change then you cut the batch in half and retest until you find the culprit.
Hope this helps someone. It seems to work for me.
Wyatt