Mindolph

v1.13 Release Note (Draft)

New, significant feature: AI Agent
- Build intelligent agents easily based on your own knowledge to enable them to answer more specialized and specific questions.
- This feature requires a PostgreSQL database (either local or remote, currently only PostgreSQL is supported) with the pgvector extension installed. The knowledge base’s vector data will be stored in this database.
- Configuration steps: Open the preferences dialog, choose the Gen-AI tab -> Configure the LLM model APIs in the Providers panel-> Configure the postgres database connection in the Options panel -> Go to the Datasets panel, create a dataset, and choose a language and an embedding model for the dataset -> Select files you want to add to the created dataset -> Click the Start embedding button to perform embedding -> After the embedding is done successfully, go to the Agents panel, create an agent, choose an embedding model and a chat model for the created agent, select your created datasets that you want to use for the created agent.
- Usage: Select the Agent tab on the left side panel -> Select an agent from the agent list that you created previously -> After the selected agent is loaded. Input something about the knowledge from the datasets in the chat box to start a conversation with the agent.

Added a global option to hide file extensions in the workspace directory tree, the recently opened files list, and the opened file tabs.
Improved the configuration of the currently used model in Gen-AI, now allowing you to select different models for both content generation and content summary.
The browser button in Mind Map note dialogs now opens URLs in an external browser by default.

Error messages from OpenAI are not displayed when generating content in a Mind Map.

Created at 2025-09-06 23:26:02

This site is open source. Improve this page.