The idea behind flagging messages as spoilers is allowing people to have a discussion without ruining the experience for others.

Mods would be able to flag a message as a spoiler, making the message hidden for the user until they click on it to reveal it.

Another idea would be a command for mods to set the channel to "spoiler sensitive" which allows viewers to self-flag their messages as potential spoilers if they are not fully sure about what is allowed to talk about or not.
Having this only allowed only whenever the channel is specifically set to this mode (for story heavy games or newly released games) would keep the abuse this function down. Also mods would be able to see the messages and could see people potentially abusing the function and then can proceed with other already given ways to deal with said person.

Example usage:

Mods usage could be
- /spoileron => "The room has been set to spoiler sensitive"
- /spoil username 2 (for the last two messages to be flagged)

Users could be
/s Has streamer X already reached part Y and beat Z?

Currently we have all kinds of ways to block various people from chatting, except the most important one: People avoiding bans by creating new accounts.

Thus we should have a new mode that only allows users with "established" accounts to chat, i.e. the account must be reasonably old.

Continuing with the theme of follower-only mode this should likely be configurable, with the number of days since an account was created for the user to be able to talk.

Suggestions for default: 7, 30, 90, and 180.

Even requiring 7 days since the account was created will make it significantly more difficult for people to do annoying ban evasion.

Follower-only mode is NOT a valid solution for this problem, as it completely ruins raids, and blocks perfectly legit users from chatting in general.

There are people on Twitch, who have a program that allows bot spamming in chats. They buy accounts somewhere in the form of login:oauth:token and connect them to their program.
They used to spam ads to promote themselves, now they do that to create conflicts between streamers or just to mess with streamers.

It is possible to get an Oauth:token through the Twitch API request, using apps verified by Twitch.
If I understood correctly from the developer's words, Twitch can get info about a token that sent the message.

What do I suggest to deal with the problem is to add a form to a channel settings in order to put there the names of users (bots), which can send messages in chat using oauth:token got by a 3rd party app.

I think it'll be the most effective way to deal with spammers because in most cases the accounts used for spamming are stolen ones and users whose accounts were stolen wouldn't be happy to find out they were banned from some channels because someone was spamming from their accounts.

This applies more to the mid to larger streamers but can be useful for the site as a whole I think.

If a person of importance is trying to communicate with the streamer then this would make it much easier for them. You may ask, why not just whisper? Well whispers aren't always suitable for those who have them disabled or for those who don't want to halt their stream to check them. The message may also be urgent so the streamer may not have time to check or may have their hands full.

This feature could also be used for mods to pin some sort of announcement in chat that viewers who just tuned in might have missed.

As for the pin feature, the broadcaster can set a maximum limit to how long these messages can stay pinned whilst moderators can pick the duration of time they want it to be pinned. If wanted, this pin option should also have the option to be disabled entirely for the viewer in their personal settings if they don't wish to see it obscuring chat. As well as this, the pinned message should have an 'x' next to it so it can be dismissed.

Provide an API hook to extension developers which notifies on each received message before it is going to be distributed to the users.

Bots can only delete already delivered chat messages afterwards. Clients can intercept the deletion. Furthermore it is not possible to write own chat commands for now.

With a hook, if a user types '/do something', extension developers were able to prevent the delivery of the message and perform some action instead.

Beyond that I would like to see API calls to send chat notifications to specific users - just like 'your message has not been sent yet. It will be reviewed by a moderator'.