Planning on making my own twitter bot.

[Algo]

I recommend to start without the auto accounts creation, instead add the ability to import accounts in a csv file.

If everything works perfectly and you need to create accounts accounts automatically then add this feature

 

[toyes88]

我目前正在使用 Selenium/Python 开发 Twitter 机器人，到目前为止一切进展顺利，我唯一的问题是：人们是否也自动执行新帐户注册过程，还是只是手动执行？
如果推文的限制率为每天 2400 条，而我需要一百万条推文，这意味着我需要大约 420 个帐户，这太疯狂了。
知道如何以正确的方式做到这一点吗？

Have you developed this software?

 

[cynicalwork]

I am currently working on a twitter bot using Selenium/Python and things are going great so far, The only question I have is : Do people automate the process of new accounts registration too or just do that manually?
If the limit rate for tweeting is 2400 a day and I need a million tweet that means I need around 420 account which is crazy.
any idea on how to do it the right way?

Hi.
As @ReverseEngineering said, you don't need selenium here, its only slowing down performance, you use a whole browser while you could simply make basic HTTP/GraphQL endpoint requests.
Yes, on twitter, people tend to automate the process of new accounts registration, because its really simple on this platform.
420 accounts on Twitter isn't crazy at all. Its really few. People are managing hundreds of thousands of accounts on this platform.

The twitter API is really easy to reverse engineer, it uses GraphQL for most of the stuff, and session storage is simple to store in a DB(ct0 and auth_token).
You don't even need to reverse engineer the real official app, and bypass SSL pinning, just intercepts requests in the browser, that is fine.

 

[OccultKingdoms]

Hi.
As @ReverseEngineering said, you don't need selenium here, its only slowing down performance, you use a whole browser while you could simply make basic HTTP/GraphQL endpoint requests.
Yes, on twitter, people tend to automate the process of new accounts registration, because its really simple on this platform.
420 accounts on Twitter isn't crazy at all. Its really few. People are managing hundreds of thousands of accounts on this platform.

The twitter API is really easy to reverse engineer, it uses GraphQL for most of the stuff, and session storage is simple to store in a DB(ct0 and auth_token).
You don't even need to reverse engineer the real official app, and bypass SSL pinning, just intercepts requests in the browser, that is fine.

Both ReverseEngineering and ekuin0x are offline.

The numbers you speak about are possible.

But it's hardcore for most people out there to just think about it.

 

[ReverseEngineering]

Both ReverseEngineering and ekuin0x are offline.

Yes, indeed I was very very busy since March 2024 (more at my profile).

But that doesn't change the fact that you won't need Selenium here. ;-)

 

[OccultKingdoms]

Yes, indeed I was very very busy since March 2024 (more at my profile).

But that doesn't change the fact that you won't need Selenium here. ;-)

Well, anything you said on the forum is unconfirmed (i.e. statement that you made $5m lifetime), so I was sceptic.

I'm still learning how to code. or more like "how to setup code so it works well with my goals"

 

[ReverseEngineering]

Well, anything you said on the forum is unconfirmed (i.e. statement that you made $5m lifetime), so I was sceptic.

Sure. But I could show you some proofs of working projects.

Please get back to me in around 2 weeks, because I have to finish another custom job first.

 

[OccultKingdoms]

Sure. But I could show you some proofs of working projects.

Please get back to me in around 2 weeks, because I have to finish another custom job first.

Well, maybe $5 million is not enough. If I made it on the internet, that probably wouldn't be enough. I can feel the pain.

 

[rampzeus]

Yes, indeed I was very very busy since March 2024 (more at my profile).

But that doesn't change the fact that you won't need Selenium here. ;-)

Although you are correct about being able to do it with requests to Twitter api and graphql.

The app itself send tens of requests anything from logging to general pings how to know what is needed and not?

How to keep the account logged in (from twitters perspective) with only requests?

 

[ReverseEngineering]

The app itself send tens of requests anything from logging to general pings how to know what is needed and not?

How to keep the account logged in (from twitters perspective) with only requests?

It's the same game like with other social media apps also:

1. Record and inspect all requests (also HTTP/2.0 and HTTP/3.0 and WebSocks should be checked).
2. Start with a small amount of requests per Twitter account.
3. Let the first bot do some random tasks for some days.
4. If the first few accounts will be locked/banned increase the number of requests and continue with #3.

 

[cynicalwork]

Although you are correct about being able to do it with requests to Twitter api and graphql.

The app itself send tens of requests anything from logging to general pings how to know what is needed and not?

How to keep the account logged in (from twitters perspective) with only requests?

I said that.
Most update requests aren't needed, focus on actions: you tweet something, then you catch the GraphQL request `CreateTweet`, and you put that into a function for your client.
An unofficial Twitter API client is less than 5k lines of codes, in most languages. Depending on the language you are using, some libraries are already there.
The account is logged in as long as the session is live, keep the same residential IP and save the `ct0` and the `auth_token`, and everything should be working fine.

However, a lot of ppl are reinventing the wheel for no meaning, like OP.

He doesn't need Selenium, nor he needs to create his own API client since the language he decided to use is Python, and there is existing open-source libraries that are here, for that language, showing that he didn't even search for them. They might not be the best option for publishing content(shadowban), but its definitely the best for scraping.
https://github.com/trevorhobenshield/twitter-api-client

And, if you don't feel like managing your own botnet, which is hard, there is paid APIs that are probably more appropriate to OP needs, since he doesn't need a lot of accounts, and he will have difficulty managing them anyway, if 420 accounts is already a lot for him. These APIs would help him getting his job done, and what he has to focus on.
https://twttrapi.com/

So, if you aren't bringing something new to the space, such as performance, or specific features, you might want to use already existing, APIs, and if you don't want to manage a lot of accounts but focus on getting the job done, paid APIs will help you as well.

3. Most apps I know limit their functions at the web interface, at least the ones I coded bots for. So the real app has more functions than the web interface. Therefore I always prefer the real app. BTW I wouldn't call the task of rebuilding web requests (browser) as "reverse engineering".

Yes, that's true, but probably not what OP is searching for, his end usage is making a bot.

Making what you are talking about would bring something to the space: higher quality accounts, less chance of getting shadowbanned. There is no limitation in terms of functions though, you can do everything on the web app.

So there is definitely meaning in creating that kind of API clients, but from a business perspective, I would sell it instead of building it myself for a simple bot.
Creating APIs through reverse engineering is really hard, requires to be technical, and take a lot of time. You need to maintain it, keep it up-to-date, migrate, fix scaling issues that comes with managing a lot of accounts.
Monetizing the API is also hard, requires to be smart, and take a lot of time. Finding clients, selling them custom traffic, custom leads, for their specific communities. Or building custom solutions and bots(what OP is doing), for their needs.
Meaning that these are two different jobs.

 

[rampzeus]

I said that.
Most update requests aren't needed, focus on actions: you tweet something, then you catch the GraphQL request `CreateTweet`, and you put that into a function for your client.
An unofficial Twitter API client is less than 5k lines of codes, in most languages. Depending on the language you are using, some libraries are already there.
The account is logged in as long as the session is live, keep the same residential IP and save the `ct0` and the `auth_token`, and everything should be working fine.

However, a lot of ppl are reinventing the wheel for no meaning, like OP.

He doesn't need Selenium, nor he needs to create his own API client since the language he decided to use is Python, and there is existing open-source libraries that are here, for that language, showing that he didn't even search for them. They might not be the best option for publishing content(shadowban), but its definitely the best for scraping.
GitHub - trevorhobenshield/twitter-api-client: Implementation of X/Twitter v1, v2, and GraphQL APIs

And, if you don't feel like managing your own botnet, which is hard, there is paid APIs that are probably more appropriate to OP needs, since he doesn't need a lot of accounts, and he will have difficulty managing them anyway, if 420 accounts is already a lot for him. These APIs would help him getting his job done, and what he has to focus on.
twttrapi - Unofficial X/Twitter API

So, if you aren't bringing something new to the space, such as performance, or specific features, you might want to use already existing, APIs, and if you don't want to manage a lot of accounts but focus on getting the job done, paid APIs will help you as well.

Yes, that's true, but probably not what OP is searching for, his end usage is making a bot.

Making what you are talking about would bring something to the space: higher quality accounts, less chance of getting shadowbanned. There is no limitation in terms of functions though, you can do everything on the web app.

So there is definitely meaning in creating that kind of API clients, but from a business perspective, I would sell it instead of building it myself for a simple bot.
Creating APIs through reverse engineering is really hard, requires to be technical, and take a lot of time. You need to maintain it, keep it up-to-date, migrate, fix scaling issues that comes with managing a lot of accounts.
Monetizing the API is also hard, requires to be smart, and take a lot of time. Finding clients, selling them custom traffic, custom leads, for their specific communities. Or building custom solutions and bots(what OP is doing), for their needs.
Meaning that these are two different jobs.

You are right.

I am doing exactly as you said, only difference is that I use the official api with the leaked twitter keys with oauth (mobile api).

I tried selenium/puppeteer before but the ram and cpu usage is very high, not even speaking about all the constant tags, id and classes that change.

How are your experiences with (shadow) bans and captcha's?

As reverseengineering said I am also doing some random actions to look more legit, I still think it would be possible to detect from them but I assume they cant log and match everything.