According to a WaPo article (which is a bit verbose to quote paras from, so I'll summarize) the key details seem to be:
Congressional appropriations of $10 million for 1999 and $31 million for 2000, although the article doesn't delve into budgetary details more than that.
It seems most of it was spent on training the opposition groups: 1st level educators trained outside Serbia, who then went back and trained a 2nd level. Also money was spent on running polls on political messaging through US firms, and even ad materials, e.g. 5000 cans of spray paint, 80 tons of adhesive paper (through USAID). The training provided was diverse, from high-level strategies like having a single, least controversial candidate, to ballot fraud detection methods (like parallel count). Election monitors were paid $5 (in a country where the monthly salary was $30 at the time); it's not totally clear if these monitor payments from the US; the article only says "Western-provided money" in that regard.
The official US line, in the words of one US NGO representative involved (the National Democratic Institute) was that the training was intended to "level the playing field". (The International Republican Institute was also involved, probably in part to make it a consensus affair in Washington.)
The trainees were instructed to avoid disclosing links to Western money. They'd give touristic reasons for visiting the neighboring countries where the training sessions were held. Some were interrogated by the Serbian police in that regard, trying to make them confess about their "Washington controllers".
According to a CNN article from 2000, there were no international observers (allowed) at those elections, so I guess training & subsidizing local observers could be justified as fair game. (The lack of access is confirmed by the OSCE; while they have a report on those elections, it starts with "The Organization for Security and Co-operation in
Europe’ Office for Democratic Institutions and Human Rights (OSCE/ODIHR) was denied entry to conduct a technical assessment and subsequently deploy an election observation mission to monitor the 24 September federal and municipal elections in the Federal Republic
of Yugoslavia. Consequently, in accordance with its mandate, the OSCE/ODIHR publishes this report of
Preliminary Findings and Conclusions without the benefit of an in-country observation mission.")