TargetWise MCP
Company and contact data for AI agents. Connect to TargetWise over Streamable HTTP to find companies and employees, enrich selected records, and retrieve available work emails and business phones.
This repository contains the official connection guide, client configurations and directory manifest for the hosted TargetWise service. The hosted server implementation and database are not distributed in this repository.
Start with one useful lookup
- Create a TargetWise account at https://targetwise.ai and obtain a workspace API key from Dashboard → Developers.
- Store the key in your client's secret store or the
TARGETWISE_API_KEY environment variable.
- Connect a client that supports Streamable HTTP and custom bearer headers.
- Use bounded search, inspect the candidates, then enrich the selected company or professional.
Customer-key clients must use /api/mcp. The hosting platform reserves /mcp; it is not the customer-key endpoint.
Claude Code
claude mcp add --transport http targetwise https://targetwise.ai/api/mcp \
--header "Authorization: Bearer ${TARGETWISE_API_KEY}"
Alternatively, use the included .mcp.json project configuration. It reads TARGETWISE_API_KEY from the environment. Keep credentials out of version control.
Cursor
In Cursor's MCP settings, configure a remote server with the endpoint and Authorization header. The included cursor.mcp.json uses Cursor's environment interpolation syntax.
Gemini CLI
Install the included extension:
gemini extensions install https://github.com/sens663/targetwise-mcp
When prompted, enter your TargetWise workspace API key. The extension declares a sensitive setting, TARGETWISE_API_KEY, and uses it in the Bearer header. Gemini CLI stores sensitive extension settings in its system keychain. Calls still consume your TargetWise workspace credits.
Clients that require an OAuth authorization flow cannot connect directly to this bearer-key gateway. Use a client that supports custom authentication headers. Do not interpret directory publication as approval by an AI platform.
| Tool | Purpose |
|---|
targetwise_find_companies | Resolve a company name, website or registration number into candidates |
targetwise_search_companies | Find companies using supported business, location, industry or technology filters |
targetwise_search_contacts | Find professional candidates at known company domains |
targetwise_search_employees | Search employee records by company, person name, job title, department or seniority |
targetwise_get_employee | Retrieve available details for one selected employee |
targetwise_autocomplete_company | Populate company fields from an identity lookup |
targetwise_enrich_company | Retrieve a profile for one selected company or supported identifier |
targetwise_enrich_contact | Retrieve requested contact fields and available professional profile context |
targetwise_find_work_email | Look up one known professional's available work email |
targetwise_find_phone | Look up one known professional's available business phone |
targetwise_reverse_email_lookup | Resolve one business email into available person and employer context |
Discovery definitions are published at https://targetwise.ai/.well-known/mcp/server-card.json. They are generated from the same tool definitions used by the running server.
Registry and discovery
The official MCP Registry record is active at https://registry.modelcontextprotocol.io/v0.1/servers/ai.targetwise%2Ftargetwise/versions/latest. Directory review and client-gallery approval are separate from registry publication.
Where it fits
- Enterprise account research: identify a company and add business context to a brief.
- Revenue operations: enrich selected records before applying your CRM's validation and merge rules.
- Contact enrichment: request an email or business phone when that field is needed.
- AI product development: ground responses in structured company and professional records.
The host owns decisions and downstream actions. TargetWise MCP does not send outreach or write to a CRM.
Usage and result handling
Data calls are metered against the workspace allowance. Consult https://targetwise.ai/pricing for current plans and result charges. Do not automatically enrich every search candidate.
- Company and contact prospecting is bounded to 25 candidates per page and 10 pages; company identity search returns at most five candidates per page.
- Field availability varies. Handle
matched, partial and not_found outcomes and preserve unresolved_fields.
- A returned phone is not necessarily a verified mobile number.
- Work-email lookup does not establish mailbox ownership or deliverability.
- Optional professional profile fields depend on account access and record availability.
- Retrieval timestamps describe when TargetWise made a request; source provenance is included only where available.
Protocol
Use Content-Type: application/json and Accept: application/json, text/event-stream. The documented compatibility flow uses protocol version 2025-11-25: initialize, send notifications/initialized, then list or call tools. Follow the live documentation for additional supported request modes.
The tools retrieve data, but calls can consume credits; consequently their annotations include readOnlyHint: false and idempotentHint: false.
Service and licensing
The MIT license covers the connection examples and documentation in this repository. Access to the hosted service and returned data is governed by TargetWise's plans and terms: https://targetwise.ai/legal/terms. The service and data are not licensed under MIT.
Privacy: https://targetwise.ai/legal/privacy
Security: https://targetwise.ai/trust/security
Support: https://targetwise.ai/company/contact