Download the PHP package cubex-hro/contao-openai-bundle without Composer
On this page you can find all versions of the php package cubex-hro/contao-openai-bundle. It is possible to download/install these versions without Composer. Possible dependencies are resolved automatically.
Informations about the package contao-openai-bundle
Contao OpenAI Bundle
The purpose of this extension is to quickly and easily generate meta descriptions and titles from page content using ChatGPT (OpenAI).
Below we have summarized a few best practices to achieve relatively good results. Additional features are planned in the future that will make the expansion even better.
Getting started
Compability
Contao Version | PHP Version |
---|---|
>= 4.9 | ^7.0 ǀǀ ^8.0 |
Important note
- An OpenAI developer account is required. Sign up here.
- The required token is also created there.
- There is a fee to use the OpenAI API. An overview of OpenAI pricing can be found here: https://openai.com/pricing
- We tested a lot and so far we haven't gotten more than $5 a month
TinyMCE Plugin notes
Please make sure you don't have a custom be_tinyMCE.html5 template. If so, take a look at src/Resources/contao/templates/be_tinyMCE.html5 and adjust the relevant places manually.
Screenshots
Best practise
- define usage limit in OpenAPI API Backend to have control over costs
How to use
- [ ] Insert token
- [ ] Choose GPT model
- [ ] Insert preferred Meta-title prompt (e.g. Schreibe für folgenden Text einen prägnanten Seitentitel bestehend aus 5 bis 6 Wörtern:)
- [ ] Insert preferred Meta-description prompt (e.g. Schreibe für folgenden Text eine informative/emphatische/ansprechende Seitenbeschreibung, die weniger als 160 Zeichen inklusive Leerzeichen enthält:)
- [ ] Set temperature (recommended: 0.5)
- [ ] Set max_tokens (recommended: 300)
- [ ] Set optional settings like hidden elements and custom fields
- [ ] Go to page settings and use the buttons below SERP preview
- [ ] Enjoy the magic :)
To-Do
- [ ] Integrate token calculator (e.g. GPT-3-Encoder-PHP)
- [ ] Content weighting through ChatGPT as pre-fetch event
- [ ] Define personality profile (role) for Chat completions API model
- [ ] Considerations and testing for the actual maximum character length for the request
- [ ] Considerations on how serialized content will be handled in the future
- [ ] Make costs per Request more transparent (show used tokens and calculate with OpenAI pricing)
- [ ] do
- [ ] some
- [ ] magic🪄
new Features
- v0.2.0 -> TinyMCE AI-Text generation Dialog
- v1.0.0 -> add Contao 5 compatibility
- v1.1.0 -> Contao Backend Help Bot powered by CustomGPT
Support
Contao Gpt Bundle is a project for the community. If you have suggestions for improvements or comments, use the issues or, best of all, make a pull request.