Two files needed to make DeepSeek work properly with Codex CLI:

model = "deepseek-reasoner"
model_provider = "deepseek-reasoner"
approval_policy = "on-failure"

The latest Paperclip release 5.1.0 has changed a little bit the way to set it up with Amazon S3 service (Amazon Simple Storage Service). Moreover, after googling a lot here and there, we could see many solutions and settings, some of them being outdated, some - often different and did not work well. So I decided to summarize in one replace all the steps needed to set up your Rails application deployed on Heroku and be able to use it with Paperclip 5 and Amazon S3 service.

In case you don't know, Heroku does not allow your Rails application to write and offers read only access. What means that you can't use Paperclip and save your files to Heroku's file system.

So you will have to find a way to upload/store/read your files. As stated in Paperclip documentation, Paperclip ships with 3 storage adapters:

Sometimes you want to use a gem on Heroku that is in a private repository on GitHub.

Using git over http you can authenticate to GitHub using basic authentication. However, we don't want to embed usernames and passwords in Gemfiles. Instead, we can use authentication tokens.

This method does not add your OAuth token to Gemfile.lock. It uses bundle config to store your credentials, and allows you to configure Heroku to use environment variables when deploying.

1. Generate an OAuth token from GitHub