---
name: privsource-expand-buyer-list
description: Use when an agent expands an existing buyer list — companies the user already has from a prior process, spreadsheet, or advisor — into net-new PrivSource buyer list results through MCP.
---

# PrivSource: Expand an Existing Buyer List

Use this workflow when the user already has buyers (from a spreadsheet, a prior sale process, or an advisor's list) and wants net-new buyers that fit the same profile.

## How existing buyers are used

- They are **positive signal**: PrivSource studies their shared industries, strategies, scale, and geography to find similar acquirers.
- They are **excluded from results**: every supplied buyer is filtered out of the generated list, including spelling, suffix, and website variants. The output is additive.

## MCP Workflow

1. Collect the existing buyers from the user. If they provide a spreadsheet or CSV, read it and extract one entry per buyer: `name` (required) and `website` (optional). Skip header, title, and total rows.
2. Call `create_buyer_list` with:
   - `description`: the company being sold (industry, revenue, geography, business model). Criteria describe the target company, not the buyers.
   - `existing_buyers`: the extracted array, capped at 1000 entries.
3. Poll `get_buyer_list_current_version` until status is `needs_review`, then confirm criteria with the user and adjust via `update_buyer_list_criteria`.
4. Call `generate_buyer_list` and read rows with `get_buyer_list_results` while status is `processing`.
5. Verify additivity if the user asks: no result row should match a supplied buyer.
6. Continue with feedback, `refine_buyer_list`, and `request_buyer_list_contact_info` as in the standard buyer list workflow.

## Requirements

- Seller account.
- Existing buyer list access enabled on the seller account. If `create_buyer_list` reports existing buyer lists are unavailable, tell the user to contact PrivSource support.

## Notes

- Passing `existing_buyers` on `create_buyer_list` is the MCP path. On the PrivSource dashboard the same feature is a CSV/XLSX upload; both feed the identical generation pipeline.
- If the user's file has unusual columns, extract the company-name and website columns yourself before calling the tool — the tool takes structured entries, not files.