MCP Server
Open tool registry, plug into Cursor or Claude
SICenter ships a native MCP (Model Context Protocol) endpoint that exposes your attack-surface data and offensive-security tooling to any compatible AI client. Connect once from Claude Desktop, Cursor, or your own agent framework and drive reconnaissance against your scoped assets without leaving your IDE.
Key capabilities
- HTTPS MCP endpoint with per-token authentication and org-scoped authorization
- Tool registry covers asset enumeration, port scanning, HTTP fingerprinting, and finding queries
- Every tool call is logged to the audit trail with the MCP token identity and timestamp
- Scope enforcement is identical to the REST API — no out-of-scope host can be targeted
- Compatible with any MCP client including Claude Desktop, Cursor, and custom agent SDKs
- Token rotation and revocation available from the dashboard without downtime
How it works
Each MCP token is bound to an organization and carries a set of allowed tool classes (read-only or read-write depending on your plan). When a client connects, it receives a tool manifest listing the available tools with their JSON Schema definitions. The client can then invoke any listed tool by name.
Tool invocations arrive at the MCP server as signed HTTP requests. The server validates the token, resolves the organization context, and dispatches the call to the appropriate backend service — the same service that powers the web console. Results are streamed back to the client as Server-Sent Events, so long-running scans return partial results as they complete.
Tokens are displayed once at creation time and are not recoverable. If a token is compromised, you revoke it from the dashboard and issue a replacement. The revocation is effective immediately; in-flight requests using the old token receive a 401 on the next event. Rotation does not require reconfiguring other tokens.
Ready? Read the MCP docs.
No credit card required. Start free, upgrade when you need more.