Add to Vector Store

Add_To_Vector_Store stages GovTribe files into an OpenAI-backed vector store so you can run semantic chunk retrieval with Search_Vector_Store.

Input Contract

  • items: Files or file-bearing entities to ingest into a vector store.

    • type: array<object>

    • required: yes

    • default: n/a

    • shape: [{"govtribe_type":"<string>","govtribe_id":"<string>"}] with 1..10 items

    • item_fields: govtribe_type (string), govtribe_id (string)

    • item_required: both govtribe_type and govtribe_id are required for every item.

    • item_notes: govtribe_type selects an eligible file/entity source and govtribe_id must be a readable GovTribe ID.

  • govtribe_vector_store_id: Existing vector store to append to; omit this field to create a new vector store owned by the caller in the active workspace.

    • type: string|null

    • required: no

    • default: null

Output Contract

  • Top-level keys:

    • response

  • Row keys:

    • n/a

  • Relationship retrieval map:

    • govtribe_vector_store_id

      • tool: Search_Vector_Store

      • filter: govtribe_vector_store_id

    • file_govtribe_id (when govtribe_type_hint indicates a user file)

      • tool: Search_User_Files

      • filter: user_file_ids

    • file_govtribe_id (when govtribe_type_hint indicates an external/government file)

      • tool: Search_Government_Files

      • filter: government_file_ids

      • fallback: if file type cannot be reliably inferred from the message line, try both tools with the same ID.

Usage Patterns

Pattern A: Create a new vector store from one known external file. Tool: Add_To_Vector_Store

Pattern B: Append new files to an existing vector store. Tool: Add_To_Vector_Store

Pattern C: Ingest contract context entities directly and let the tool resolve their attached files. Tool: Add_To_Vector_Store

Last updated

Was this helpful?