Greenhouse Recruiting Client Credentials auth, new OpenAI model, and other app updates

here's a quick look at our app updates this week greenhouse recruiting new client credentials authentication you can now connect to greenhouse recruiting using the oauth 2 0 client credentials flow this method authenticates directly from make's servers with a client id and client secret, so scenarios don't need a user based login and consent screen the new connection type works with all action and search modules, but not with instant trigger (watch events) modules; use the original greenhouse harvest connection type for webhooks for more information, see greenhouse recruiting docid\ inj0xb7xfdui7joo1njaz page gotowebinar webhook creation fix we've resolved an issue where creating a webhook for the webinar updated event in gotowebinar trigger modules failed you can now create these webhooks successfully gotowebinar enhanced connection security we've updated the gotowebinar app to better protect your connection credentials the client secret field is now masked by default when you create or edit a connection, and you can view it using the show icon this change doesn't affect existing connections vonage enhanced connection security we've updated the vonage app with security enhancements for your credentials the api secret is now masked upon entry, and authorization tokens are sanitized to prevent accidental exposure wordpress fix for the create a post module make resolved an issue where the create a post module could complete without an error but fail to create a post in wordpress, typically when a cdn or caching service was in use the module now returns an error when it doesn't receive a post id, with guidance for troubleshooting cdn configurations openai new gpt 6 astra model available make added support for gpt 6 astra, openai's latest model, across openai modules updated modules simple text prompt you can now select the new gpt 6 astra model from the model dropdown please note that this model is available on paid make plans generate a response (and other modules) the gpt 6 astra model is now available in modules with dynamic model selection it appears under a dedicated "gpt 6" group for easy access for more information, see openai modules docid\ erxxzx7ntkonc19sojho8 openai enhancements to the create a chat completion module make improved how the create a chat completion module handles the tool choice parameter when you select none , the tools (functions) field is now hidden, and the module no longer causes an error if no tools are defined selecting auto or required will now correctly mark the tools (functions) field as mandatory openai new vector store and embedding modules make added modules for managing vector stores and generating embeddings in the openai (dall e & gpt 3/4) app, so scenarios can use semantic search and long term memory new modules generate an embedding creates vector representations of text for use in similarity searches, clustering, and other machine learning tasks list vector stores retrieves a list of your available vector stores search a vector store finds the most relevant text chunks within a vector store based on a natural language query delete a vector store permanently removes a specified vector store updated modules create a vector store or add files you can now create an empty vector store without uploading files at the same time this module also includes a new chunking strategy parameter for more control over how your files are processed for search for more information, see openai modules docid\ erxxzx7ntkonc19sojho8 openai bug fixes for the upload a file module upload a file module the module's user interface has been updated to align with recent openai api changes the "purpose" field now has a clearer "file search / vector store" label instead of the outdated "assistants" label, and new "user data" and "evals" options have been added help links in the module have also been updated to point to current documentation upload a file module an issue has been resolved where the module's output for the created at field was an unformatted unix timestamp this field is now correctly parsed as a date