mcp inspect

    1

    Inspect, debug, and call MCP servers from the CLI with support for stdio, HTTP, SSE, and complex auth.

    Free

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    mcp inspect

    Example session with this skill installed

    Show me the tools available on the local knowledge base server running via npx. The command is 'npx -y kb-server'. Also show the schemas.

    • Read your context and instructions
    • Compiled the mcp inspect

    I have inspected the tools and their respective input schemas for the kb-server using the mcp-inspect CLI.

    MCP Tools: kb-server

    ToolDescriptionArguments Schema
    searchSearch the knowledge base for a substring.{"type":"object","properties":{"substring":{"type":"string"}},"required":["substring"]}
    readRead a specific document by ID.{"type":"object","properties":{"id":{"type":"string"}},"required":["id"]}
    list_docsList all document IDs and titles.{"type":"object","properties":{},"required":[]}
    [
      {
        "name": "search",
        "description": "Search the knowledge base for a substring.",
        "inputSchema": {
          "type": "object",
          "properties": {
            "substring": {
              "type": "string"
            }
          },
          "required": [
            "substring"
          ]
        }
      },
      {
        "name": "read",
        "description": "Read a specific document by ID.",
        "inputSchema": {
          "type": "object",
          "properties": {
            "id": {
              "type": "string"
            }
          },
          "required": [
            "id"
          ]
        }
      }
    ]
    

    Next steps
    1.

    Test a tool call: Run mcp-inspect call search -- npx -y kb-server --args '{"substring":"test"}' to verify the output.
    2. Save the server: Use mcp-inspect add kb -- npx -y kb-server to avoid typing the full command each time.
    3. Verify result format: Pipe the output of a call to jq to see how the server structures its content field.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Inspect, call, and debug MCP servers from the command line, without writing client code.

    What it does

    • Server info: name, version, protocol version, capabilities, and instructions.
    • Tool schemas: lists tools and shows each tool's parameters (types, defaults, constraints, nested objects) and result schema.
    • Tool calls: calls a tool with JSON arguments (inline, from a file, or from stdin) and prints the raw result. Exit codes distinguish connection errors from tool errors.
    • Transports: stdio, streamable HTTP, and legacy SSE.
    • Auth: OAuth (discovery, dynamic client registration, PKCE, refresh), static bearer tokens, tokens from a command, custom headers.
    • Debug output: three verbosity levels, covering auth and transport decisions, JSON-RPC messages, and raw HTTP traffic with secrets redacted.

    How it works

    1. Pick a target: a saved server name, a URL, or any command that starts a stdio server (e.g. -- npx -y some-server).
    2. Inspect: info, tools, and tool <name> show what the server offers.
    3. Call: call <tool> --args '{...}' runs the tool and prints the JSON result.
    4. Debug: -v to -vvv show where a connection or login fails.

    Built with

    Go, on the official MCP Go SDK. Works with any MCP server.

    Use cases

    • Check what a new MCP server exposes and whether its schemas are what you expect.
    • Debug connection and authentication failures on remote MCP servers.
    • Call MCP tools from shell scripts, with jq for the output.

    Known limitations

    • Requires Go 1.26 to install (or Go 1.21+ with GOTOOLCHAIN=auto).
    • An agent can't complete the OAuth browser login. Log in once yourself, or use a bearer token.
    • Supports tools only, not resources or prompts. No sampling, elicitation, or roots.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Testing & Debugging.

    Requires Go 1.26+ (or 1.21 with GOTOOLCHAIN=auto). Agents cannot complete interactive OAuth flows; users must run the add command manually for initial browser-based auth.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free to use

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 16 days ago

    • Free with an Agensi account

    Software
    Go 1.26Some MCPNode.js

    Permissions

    Terminal / Shell
    Read Files
    Network Access
    Listed16 days ago

    What's inside

    Frequently Asked Questions

    This CLI tool eliminates the trial-and-error process of debugging MCP servers by allowing you to inspect schemas, test tool execution, and monitor raw JSON-RPC traffic without writing custom client code.

    The tool is compatible with any server following the MCP standard and supports stdio, HTTP, and SSE transports, making it ideal for debugging both local and remote agent integrations.

    Yes, it includes advanced authentication handling for static bearer tokens, command-based token generation, and OAuth 2.0 flows, helping you troubleshoot connection issues in secure environments.

    You will need Go 1.26 or higher installed on your system; access to network and ability to run "go install"; once configured, you can point the tool at any MCP-compliant server to begin immediate inspection and testing.

    While the tool handles the protocol-level handshake, interactive OAuth 2.0 flows require a one-time manual setup by the user, as an automated agent cannot complete external browser-based authorizations.

    The skill includes instructions to install the tool, comprehensive documentation for all commands (info, tools, call), and three levels of verbose logging for deep debugging.

    New in Testing & Debugging

    Free