Access and authentication
What is the FullHunt MCP server?
It is a hosted Model Context Protocol server that maps FullHunt REST API operations to tools. An MCP client can call those tools and combine their responses in a conversation or automated workflow.How many tools are available?
The current server defines 63 tools. The Tools Reference is organized by account, passive intelligence, vulnerability, Enterprise, Nexus, OEM, and public operations.Which tools can my account use?
Availability depends on the plan and modules associated with the API key. Callauth_status first, then consult Plans, Access, and Credits. A tool can appear in the client while its underlying API still returns 403 Forbidden for an account without access.
How is the server authenticated?
The hosted endpoint expects the API key in theX-API-KEY header. Do not use a bearer token unless a specific future integration explicitly documents bearer authentication.
Client setup
Does FullHunt work with Cursor?
Yes. Configure the hosted URL andX-API-KEY header in .cursor/mcp.json or ~/.cursor/mcp.json. See the integration guide.
Can I add the hosted server to claude_desktop_config.json?
Do not use the legacy local-server configuration for this hosted endpoint. Claude manages remote integrations through Settings > Connectors. FullHunt’s hosted endpoint requires an API-key header, so direct use depends on whether the connector flow available to your account supports that authentication method.
What is the best connection test?
Callauth_status. The old get_my_ip verification prompt is invalid because the MCP server does not expose a get_my_ip tool.
Tool behavior
Is fullhunt_host a live scan?
No. It retrieves indexed host details. Use fullhunt_scan, enterprise_on_demand_scans, or oem_on_demand_scan only when you need an active scan and have authorization.
Can MCP tell me whether an IP is malicious?
Nexus tools provide fields such as IP context, Tor status, passive DNS, and reverse-host relationships. They do not provide a universal binary verdict. Ask the client to separate returned evidence from its own analysis.Can MCP schedule monitoring or send alerts?
The FullHunt tools return data and can trigger documented scan operations. Scheduling, notifications, tickets, and report storage must be implemented by the MCP client, a workflow engine, or your application.How do search and fetch work?
search classifies a domain, host, CVE, organization, or free-text query and returns retrieval stubs. fetch resolves a selected stub through the relevant FullHunt operation. These tools support clients that use search-and-fetch research patterns.
How do I check an OEM scan?
Calloem_on_demand_scan, retain the returned scan_id, and poll oem_scan_status. This status tool is OEM-only. Standard and Enterprise scan tools do not expose a matching MCP status operation.
Data and safety
Is every response current?
No. Most tools query indexed datasets with source-specific collection times. Use returned timestamps where available and review Data Provenance and Freshness.Can I scan any target?
No. Active scanning is limited to assets you own or are explicitly authorized to test. Reserved example domains and IP ranges illustrate syntax only and must not be treated as scan targets.Does FullHunt store my MCP conversation?
The MCP server receives tool arguments and returns tool results. Client-side conversation retention is controlled by the MCP client. Do not include secrets or unnecessary personal data in prompts or tool arguments.Troubleshooting
Why do I receive 401 Unauthorized?
The X-API-KEY header is missing or invalid. Verify the same key against GET /api/v1/auth/status.