Protocol Standards Integration
Model Context Protocol (MCP) Consent & Authorization Layer
Secure MCP server tools and resource prompts with human consent screens.
Integration Code Example (TYPESCRIPT)
import { Server } from "@modelcontextprotocol/sdk/server/index.js";
import { Permitly } from "permitly";
const permitly = new Permitly({ apiKey: process.env.PERMITLY_KEY });
server.setRequestHandler(CallToolRequestSchema, async (request) => {
const consent = await permitly.requireConsent({
scope: `mcp.tool.${request.params.name}`,
params: request.params.arguments
});
return { content: [{ type: "text", text: "Authorized action executed." }] };
});Execution Sequence Flow
sequenceDiagram
participant LLM as 🧠 LLM Client (Claude/Cursor)
participant MCP as 🔌 MCP Server
participant Permitly as 🛡️ Permitly Layer
participant User as 👤 Human Approver
LLM->>MCP: 1. CallToolRequest (mcp.tool.filesystem)
MCP->>Permitly: 2. Check Consent Token
Permitly->>User: 3. Render Hosted Consent Screen
User-->>Permitly: 4. Approve Request
Permitly-->>MCP: 5. Return Validated Token
MCP-->>LLM: 6. Return Tool Execution Result
High-ROI Security Gate
Protecting your Model Context Protocol (MCP) pipeline against unauthorized actions delivers an immediate compliance audit trail with zero infrastructure overhead.
Frequently Asked Questions
Is Permitly compatible with standard MCP servers?
Yes, Permitly integrates with any TypeScript or Python MCP server implementation via standard request handlers.