HubSpot via Windsor.ai

Windsor.ai exposes live HubSpot CRM and marketing data (read only) through MCP tools. The connector id is hubspot. Pass hubspot as the connector on every tool call; the Windsor.ai server serves all connectors, so it is never implied.

Golden rules:

• Never guess field, account, or option names. They come from get_fields, get_connectors, and get_options. Call those first; do not invent identifiers.
• HubSpot is read-only here — there are no write actions. Do not offer to create, edit, or delete records, or to send emails.
• Report the numbers the user asked for, concisely. Don't dump raw rows unless asked.

1. Ground yourself first

Call get_connectors to see which HubSpot accounts are connected (each has an id and, usually, a name). If none are connected, help the user connect one: call get_connector_authorization_url (or get_connector_connect_info for the auth type and steps) and give them the setup link — do not describe manual dashboard navigation.

If the user names an account, match it to a connected account id. If several are connected and the request is ambiguous, ask which account.

2. Reading CRM and marketing data

1. Call get_fields for hubspot to get valid field ids — contacts, companies, deals, deal stage, pipeline, deal amount/value, close date, owner, lead source, and marketing metrics (email opens, clicks, conversions, revenue). Use the exact ids returned, not these examples.
2. If the analysis needs segmentation or attribution controls, call get_options to see the connector's options and valid values.
3. Call get_data with:
   • fields: the ids you actually need (keep the set small).
   • accounts: the target account id(s); omit for all connected accounts.
   • A date range: date_from/date_to ("2026-07-01"), or a date_preset (last_7d, last_30d, last_90d, this_month, this_year, last_year; append T to include today).
   • options / filters only if relevant, e.g. [["deal_stage","eq","closedwon"]].

Common requests → recipe:

• Pipeline value: fields deal_stage + deal amount; aggregate by stage.
• Win analysis: filter to closed-won deals, group by lead_source or owner.
• Marketing performance: email/campaign fields (sends, opens, clicks, conversions) over a preset range; compute open and click rates.

Watch out: money fields are in the account currency (state it if known). If a field the user wants isn't in get_fields, say so rather than substituting a different metric.

3. Reference the user's data, don't fabricate

If a tool returns an error (unknown field/connector/account), call the relevant discovery tool (get_fields, get_connectors) and retry with correct ids. Never fill gaps with made-up numbers, deal names, or contacts.