Inspect, validate, debug, and sanitize Model Context Protocol (MCP) JSON-RPC 2.0 requests, responses, and notifications for Claude Desktop, Cursor, Zed, and WebMCP. 100% client-side privacy ensures sensitive tool parameters and API keys never leave local browser memory.
Compliant (`jsonrpc: '2.0'`)
Request Frame (method: 'initialize', id: 1)
Valid ID (number: 1)
Protocol version: '2024-11-05'
No unmasked API keys or bearer tokens detected.
Select a starter preset (`initialize`, `tools/list`, `tools/call`, `resources/read`) or paste your custom MCP JSON-RPC 2.0 message payload.
Inspect JSON-RPC 2.0 compliance, method parameters, and error codes. Click auto-redact to scrub hardcoded secrets (`sk-`, `ghp_`, Bearer tokens).
Copy ready-to-run TypeScript (`@modelcontextprotocol/sdk`) or Python (`mcp`) client integration code and inspect sequence diagrams.
Zero backend network calls. All MCP JSON-RPC 2.0 frames, parameters, system messages, and API tokens are evaluated strictly inside local browser memory.
Validates `jsonrpc: '2.0'`, message ID types, method names, params schemas, result structures, and standard error codes (`-32700` to `-32603`).
One-click starter frames for Server Initialization (`initialize`), Tool Discovery (`tools/list`), Tool Execution (`tools/call`), Resource Reading (`resources/read`), and Notifications.
Scans payload strings for unmasked credentials (`sk-`, `ghp_`, Bearer tokens) and automatically redacts them to prevent token leaks.
Export copy-paste client and server message handling code for TypeScript (`@modelcontextprotocol/sdk`) and Python (`mcp`).
Renders an interactive message sequence flowchart illustrating Client <-> Server request-response cycles and error recovery flows.
Our MCP Inspector is open source under HiMat Technologies. Star the repo or submit improvements on GitHub.
HiMat's MCP Inspector operates 100% locally in client browser memory. Neither your tool arguments, database queries, file paths, nor API credentials are ever sent to HiMat or any third-party servers.
It is a developer utility that parses, validates, and debugs Model Context Protocol (MCP) JSON-RPC 2.0 request and response frames exchanged between AI clients (like Claude Desktop, Cursor, Zed, or WebMCP) and MCP servers.
No. The HiMat MCP Inspector operates 100% locally in your web browser. Neither your JSON-RPC frames, tool arguments, local file paths, nor API credentials leave your device.
Our inspector supports all standard MCP methods including `initialize`, `notifications/initialized`, `tools/list`, `tools/call`, `resources/list`, `resources/read`, `prompts/list`, `prompts/get`, and custom notification events.
Our built-in local scanner inspects string fields in params and result objects for common API token signatures (e.g., `sk-`, `ghp_`, `bearer`) and replaces raw values with `[REDACTED_API_KEY]`.
Yes, 100% free with no account registration required, no daily rate limits, and zero advertisements.
HiMat Technologies builds custom AI agent architectures, multiagent orchestration pipelines, tool-use APIs, and 21-day fixed-scope AI pilots with complete ownership transfer.
Continue with related utilities, services, and guides from HiMat.