Dataset Operator

[sdflkjssl]

Self Checks

• I have searched for existing issues search for existing issues, including closed ones.
• I confirm that I am using English to submit this report (我已阅读并同意 Language Policy).
• [FOR CHINESE USERS] 请务必使用英文提交 Issue，否则会被关闭。谢谢！:)
• Please do not modify this template :) and fill in all the required fields.

Content

Are there dataset operators in the open source version? If no, how can I achieve similar functionality?

[dosubot[bot]]

Hi @sdflkjssl! I'm Dosu and I’m helping the dify team.

There are no dataset operators as a distinct feature in the open source version of Dify. However, you can achieve similar functionality using Dify's knowledge pipelines, dataset APIs, and UI components. Dataset management tasks—like selection, embedding, file upload, document handling, error retries, and cleanup—are handled programmatically and through the UI, not via a dedicated "operator" abstraction.

You can define and orchestrate data processing using the knowledge pipeline DSL (YAML-based), interact with datasets via API endpoints, and use custom scripts for advanced operations. Role-based access (e.g., dataset_operator, dataset_editor) controls permissions for dataset actions, but these roles do not provide a general-purpose operator feature. Some tasks, such as re-uploading documents or cleaning up vector collections, may require manual intervention or scripting, especially when dealing with schema changes or database migrations. For more, see the docs on knowledge pipelines and recent dataset API changes.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

   

[sdflkjssl]

@dosu I want a user to be only able to edit the knowledge base but not the application. How can I do that?

[dosubot[bot]]

You can achieve this by assigning the DATASET_OPERATOR role to the user. This role allows users to edit the knowledge base (datasets) but does not grant them permission to edit applications. Only OWNER, ADMIN, and EDITOR roles can edit applications, while DATASET_OPERATOR is limited to dataset management tasks. You can set this role for a user in your workspace via the admin UI or API. There is no more granular RBAC in the open source version—roles are fixed and permissions are enforced in the backend code.
See code reference

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

   

[sdflkjssl]

@dosu Where is the admin UI in the open source version? I can't find it.

[dosubot[bot]]

There isn't a separate "admin UI" page in the open source version. User and role management (including assigning roles like DATASET_OPERATOR) is handled in the workspace/account settings modal. To access it, click your avatar in the top-right corner of the Dify web app and select "Workspace settings" (or similar). In the modal that opens, go to the "Members" tab—this is where you can invite users, assign roles, and manage workspace members. This modal is effectively the admin UI in the open source version.

Note: In the self-hosted open source version, only one admin user is supported due to licensing; multi-admin (multi-tenancy) requires a business license. Some settings and tabs may be hidden depending on your role (for example, DATASET_OPERATOR sees fewer options) citation.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}