Creating glossaries with the MCP create_glossary_from_files tool

Automate the creation of structured business glossaries from PDF and XSLX documents that you upload to a project. The MCP create_glossary_from_files tool extracts business terms from these documents, generates descriptions, establishes relationships, organizes terms into categories, and creates a structured glossary for you.

Watch this video to learn how you can turn your business documents into structured glossaries with the create_glossary_from_files tool.

This video provides a visual method to learn the concepts and tasks in this documentation.

User permissions

  • Access catalogs
  • Access data preview
  • Access governance artifacts
  • Manage governance artifacts
  • Create projects

Supported file formats for generating glossaries

You can upload files in these formats:

  • PDF
  • XLSX

To upload the files, on the project Assets page, click Import assets, select the Local file type and add your file. All documents that you add to the project are processed incrementally.

Consider the following tips:

  • Upload files such as business documents, policies, procedures, industry standards. Select documents with clear business language rather than purely technical code without business context to generate a well-structured glossary.
  • Split very large documents into logical sections. Multiple smaller files are processed faster.

Prompting the AI agent

Prompt example: Create a glossary from files for project: *Project_name* id: *Project_ID*.

Consider the following tips:

  • Include both the project name and project ID for accuracy.
  • If you don't know the project ID, you can ask your agent to list it for you.

create_glossary_from_files workflow

  1. Upload a document or documents to a project.
  2. Prompt the AI agent to create a glossary from file in the project.
    The tool runs in 3 different modes. After each mode completes, you can review the output and decide to proceed with the next mode.
    You can add and delete documents in the project and run the Process and Review modes again.
  3. The tool processes the documents (Process mode).
    Candidate terms are extracted and a corresponding .json file with the processed data is created in the project. The ..json file is updated incrementally after each document is processed.
  4. The tool reviews the .json file and generates a glossary structure (Review mode).
    You provide a name for the parent category. The extracted terms are analyzed. Non-data related terms are filtered out. Descriptions, categories, and relationships are generated. The complete glossary structure with categories and business terms is created in two CSV files in the project. You can review the returned glossary statistics and confirm to create the glossary structure in your project.
  5. The tool applies the glossary structure in the project (Apply mode).
    The categories and draft terms are imported. You can use the provided links to open to the parent category in the UI and review the terms.

Learn more