---
name: privsource-buyside-mandates
description: Use when an agent creates, updates, archives, or publishes the authorized buyer account's own PrivSource buyside mandates through MCP, or reviews the deals PrivSource matched to them.
---

# PrivSource Buyside Mandates

Buyside mandate tools are buyer-side. They manage the authorized buyer account's own acquisition mandates — the criteria sellers browse when looking for buyers — and the deal matches PrivSource finds for published mandates. Drafts are private; publishing makes a mandate visible to seller accounts.

## Workflow

1. Call `privsource_whoami` if unsure whether the account has `buyside-mandates` access.
2. Call `list_buyside_mandates` to find existing mandates before creating a new one. Filter with `status` (`draft`, `published`, `archived`).
3. Call `create_buyside_mandate` with a title to start a draft. Add criteria in the same call or later with `update_buyside_mandate`.
4. Call `get_buyside_mandate` and check `publishable` and `missing_fields` before publishing.
5. Call `publish_buyside_mandate` only when the user asks to publish and `missing_fields` is empty. Publishing makes the mandate visible to sellers, so confirm intent first.

## Reviewing Deal Matches

PrivSource matches live deals against published mandates. Each mandate from `list_buyside_mandates` includes `match_count`, `strong_match_count`, and `to_review_count` — use them to pick which mandate has matches worth reviewing.

1. Call `get_buyside_mandate_matches` with the mandate `id`. Matches come back ordered by fit with a `score`, a `fit_band` (`strong` or `possible`), an `explanation`, and a `review_state`. Filter with `review_state` (`to_review`, `in_pipeline`, `not_a_fit`); `meta.review_state_counts` shows the split.
2. If `meta.matches_first_time_finding` is true, PrivSource is still building the mandate's first matches — say matching is in progress rather than reporting that nothing matched.
3. Each match includes a deal summary. `available: false` means an engaged stale match remains visible for history but the deal is no longer on the market. `direct_listing: true` means the seller shared it directly; `shared_by` identifies that seller, and Direct Listings lead their fit band.
4. Pass an available match's `deal.id` to `get_deal` for full details; `send_deal_outreach` works on contactable deals.
5. Call `update_buyside_mandate_match` to record the buyer's verdict: `review_state: in_pipeline` saves the deal to the buyer's pipeline (Interested on a direct listing), `not_a_fit` passes on the deal and hides it from the feed, and `to_review` undoes either. Set `feedback_comment` only for `not_a_fit` matches to explain why you passed.
6. Saving to pipeline and hiding act for the whole buyer account, so confirm intent before changing review states in bulk.

## Mandate Learnings

When the account has mandate learnings, `get_buyside_mandate` includes `learnings` and `archived_learnings` arrays from the buyer's match feedback. Each learning has these fields:

- `id`: the identifier for `update_buyside_mandate_learning`.
- `dimension`: the observable deal attribute (`sector`, `geo`, `size`, `business_model`, `customer`, or `characteristic`).
- `direction`: the ranking effect (`exclude`, `penalize`, or `boost`).
- `rule`: one declarative matching rule.
- `evidence_type`: `explicit` for comments or `implicit` for behavior.
- `event_count`: the number of supporting feedback events.
- `status`: `pending` for a one-event signal or `active` after two or more supporting events.
- `observations` and `evidence_match_ids`: the evidence summary and referenced matches.

`learnings_analyzing` is true while a refresh is running. Treat a pending learning as a weak signal.

Review active learnings before editing a mandate. Apply agreed changes with `update_buyside_mandate`.

Ignore a learning with `update_buyside_mandate_learning` and `archived: true`. Restore it with `archived: false`. These are the same actions as the dashboard.

## Field Meaning

All criteria describe the companies the buyer wants to acquire.

- `min_revenue` / `max_revenue` and `min_ebitda` / `max_ebitda`: target company financials in whole US dollars. Use the `no_min_*` / `no_max_*` booleans when the mandate explicitly has no bound.
- `industry_groups`: the primary `industry` must come from the PrivSource mandate taxonomy (enumerated in the tool schema); each group takes optional `sub_industries` and a free-text `niche`. The array replaces the mandate's existing groups.
- `locations`: PrivSource location names such as countries, U.S. regions, U.S. states, or Remote. The array replaces the mandate's existing locations.
- `industry_agnostic` / `no_geographic_preference`: set these instead of industries/locations when the mandate has no preference.
- `description`: the investment thesis; must be at least 100 characters before publishing.
- `close_timeline`: required before publishing.

## Publish Requirements

Publishing requires a descriptive title, a 100+ character description, industry preferences (groups, agnostic, or exclusions), at least one revenue or EBITDA value, geography (locations or no preference), and a close timeline. Resolve `missing_fields` from `get_buyside_mandate` first.

## Error Handling

- Archive and unarchive with `update_buyside_mandate`'s `archived` flag. Archived mandates must be unarchived before publishing.
- If a tool reports validation errors, fix the named fields and retry rather than guessing new values.

Feedback notes are only available for `not_a_fit` matches. Set the state and pass reason together, or edit an existing pass reason.

## Exports

Call `get_buyside_mandate_matches_export` with the mandate `id`.
The response returns the file directly in `export.content`. No browser sign-in is needed.
If `encoding` is `base64`, decode `content` to bytes. If it is `utf-8`, use the text directly.
Save the result using `filename` and `content_type`, or analyze the CSV text in the agent.
Check `ready` first. Unavailable exports return `unavailable_reason` without file content.
The CSV includes all dashboard review states and fixed columns. Paid buyer and Deal Network access are required.
`dashboard_url` is optional context. It is not required to get the file.