I just finished implementing an experimental search engine for ANN. Even though I implemented it in true trial&error, quick&dirty (TEQD) fashion - copying various parts of ANN's existing code and changing things around until it gave the desired results - it should work fairly well. Speed was excellent, considering it does a full text search of a few thousand files. How it works is you enter a single word (multiple words, and/or cases, or upper/lowercase distinction are all not supported yet), select in which category or categories you want to search, and whether you just want to check the titles (headlines) or the full article. Note that in some cases a word might only be found in the "Read More" part, which is not displayed right away, but you have to click on it like you usually do in such cases.
Please let me know if you encounter any bugs or inconveniences, and do not hesitate to add any other ideas and suggestions in the comments section.