> For the complete documentation index, see [llms.txt](https://cleyrop.gitbook.io/docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://cleyrop.gitbook.io/docs/documentation-fr-en/support-and-resources/changelog-and-release-notes/latest-release-note.md). # Latest Release Note ## 📣 Announcements Version 4.6 enhances Le Druide: you can now generate office documents, with iteration through comments, and create agents directly from the interface. The Agents’ knowledge scope can now include datasets. Data connectors are also expanded with two new connectors dedicated to Business Central and Salesforce, and the API connector now handles generic pagination. Merge conflicts on Dataflows can now be resolved directly from the merge request. *** ## 🚀 New features ### Le Druide ### Office document generation * Generate office documents (Word, PDF, PowerPoint, Markdown) from the global assistant, from a natural-language prompt. * Preview the generated document before downloading it. * Iterate on a generated document by commenting on it: the assistant produces a new version taking your feedback into account. → [Generate office documents](/docs/documentation-fr-en/le-druide/generate-documents.md) ### Create an Agent from Le Druide * Create personal agents directly from Le Druide, without going through AI Studio. * These agents are personal and cannot be shared. → [Using Agents and Tools](/docs/documentation-fr-en/le-druide/utiliser-les-agents-et-les-tools.md) ### Cleyrop Dataset - Download the result of a query on a dataset * Query a dataset you have access to by mentioning its techname: Le Druide runs a SQL query and displays its execution in an expandable block. * The query result can be downloaded as a file. → [Query datasets](/docs/documentation-fr-en/le-druide/interroger-des-datasets.md#telecharger-le-resultat-dune-requete-sur-un-dataset) ### AI Studio: Add datasets to Agents’ knowledge scope * Associate one or more project datasets with an Agent’s knowledge scope. * Adding a dataset automatically activates the required MCPs, with no manual configuration. → [Create an Agent](/docs/documentation-fr-en/data-and-ai-project/studio-ai/agents.md#creer-un-agent) *** ### Dataset - extension of supported types * The supported types for datasets have been expanded: Decimal, Timestamp, Binary, typed Array, Struct/Map. → [Dataset Types](/docs/documentation-fr-en/data-and-ai-project/datasets/gerer-les-datasets.md#types-de-colonnes-supportes) *** ### Data sources — New connectors * Connect to **Business Central** and **Salesforce** via dedicated connectors, with simplified authentication. * The generic API connector now supports **generic pagination** (page, offset, cursor, link…), in addition to the OData protocol. → [Data sources](/docs/documentation-fr-en/data-governance/sources-de-donnees.md) *** ### Dataflow — Checking and resolving conflicts * Check the result of the previous step while building a Low Code transformation, using the **Check previous step**. * Resolve merge conflicts directly from the merge request view, for Python or SQL transformations. → [Collaborate and version](/docs/documentation-fr-en/data-and-ai-project/dataflow/collaborer-et-versionner.md) *** ### The Factory * Add multiple environment variables at once by pasting them directly into the dedicated field. * Open additional TCP ports on your apps (`22` for SSH, `443` for HTTPS). * Reference an image by its SHA256 digest (`image@sha256:`), in addition to the tag format, to guarantee an immutable reference. * Cleyrop’s native Tools (Corpus, Dataset, Internet) are now visible in the Apps and Tools list. → [Manage instances](/docs/documentation-fr-en/the-factory/gerer-les-instances.md) *** ## 🔧 Other changes and improvements {% hint style="warning" %} Security hardening — SFTP authentication → [Data sources](/docs/documentation-fr-en/data-governance/sources-de-donnees.md#depots-sftp-securises) * Creating new repositories or password-based SFTP access is no longer possible: only SSH key authentication is now offered, in line with ANSSI/NSA recommendations. * Existing repositories configured with a password remain accessible, and their password can still be reset. {% endhint %} * **Cleyrop Dataset** benefits from several improvements: the limit on returned columns is removed, datasets can be filtered by classification, and dataset name resolution has been improved. * **Dataflow Cluster :** Restart the build of a custom cluster image using the Rebuild button — useful, for example, to retrieve the latest available versions of libraries without a pinned version, without having to recreate the image. * **Dataflow** : Reference multiple output datasets at once when deploying a Dataflow, with a Select all option. * **Cleyrop Dataset** benefits from several improvements: the limit on returned columns is removed, datasets can be filtered by classification, and dataset name resolution has been improved. * **The Factory:** Container logs are now exposed when an App deployment fails, to make troubleshooting easier. * **Administration** : Platform Managers can now delete a DevSpace from Administration. * **Dataset API** : Several optimizations improve the performance of the Data Serve API. * **Corpus API** : A link to the Corpus API Swagger documentation is now available. * **AI Studio** : Creating new AI Assistants is now blocked in anticipation of their upcoming decommissioning, in favor of Agents. *** ## Bug fix * Fix for the agent creation error in AI Studio after reactivating a user. * Fix for an error when deleting a conversation history on Le Druide. * Fix for selection in the trash. * Unsupported files (e.g. JSON) now display a clear message when creating an agent corpus. * Fix for the button that opens an app in the Applications list. * Fix: the connection test to an API datasource did not actually verify access. * Fix: deleting and recreating a DevSpace could cause the SSH key to be rejected on Gitea.