No one ever actually used any of the IRC bots, so one by one they have gone offline, I am not even sure exactly why since I didn't actually turn off their entries in the crontabs of the various usernames running them. I think eggdrop just tends to eventually corrupt its user file, as while I did care abou them I did have to restore the user files of several of them to keep them working. Quite likely since I stopped caring they have one by one corrupted their user files (which, since they have no users, mostly contains their entries about each other so they can network among themselves).
That corruption isn't a dealkiller since part of the purpose of their networking among themselves is to share their users, so as long as you can find one still running it should know about you thus let you transact exchanges between the various blockchain-based currencies.
However, I never did get the bot I tried to build OT's TCL API into to work. It has turned out in the intervening year that actually OT needed a lot of work, we found some deep problems and fixed them but by then the market had made it clear that IRCbot-mediated exchanges were not wanted/desired.
My focus has thus, in that intervening year, moved on to running an actual Open Transactions server, and the debugging and so on has been focussed on getting that, and a demo client demonstrating how one could talk to such a server, up and running and, recently, also able to be built using autotools and having windows binaries ready to download.
I have a thread in the projects section about my Digitalis Open transactions server
which if nothing else may serve to chronicle some of what has been done during the last year or so.
User balances have survived admirably, they are very robust, the system works well in that respect. The last few months have mostly been focussed on implementing paranoid security so that a user's machine never lets the passphrase used to decrypt the user's private keys get written to disk. Right now custody of an intermediate key derived from the passphrase is in the process of being delegated to the "keyring" services of the various supported operating-systems so that even that keyring does not store the actual passphrase, just something derived from it that Open Transactions can use to decrypt the user's private keys for a configured number of seconds. It has taken a lot to get this far with that degree of paranoia, and all the constant testing implemented by automatically run scripts has halted ever since the paranoia reached a point where the scripts could no longer work unless a human actually "held their hand" every time they run, providing the passphrases over and over again at all times of day and night when the scripts got run by the scheduled-tasks service (cron, as I use Linux).
The IRC bots that actually worked did not use Open Transactions; they simply called shell scripts that communicated with *coin daemons, using the verified "nick" of the IRC user as the "account" name in the wallets controlled by the *coin daemons; and actually I do not really know whether they did really work as no one evidently ever managed to create an account on one, albeit only one or two attempts that I know of were claimed to have been made, in neither case persistently enough to proceed to an actual test/exploration to try to figure out what their problem had been.