DalšíDalší průvodce
Použití LLM Playgrounds k testování výzev
Technický
Technický PRŮVODCE
MCP Inspector je vývojářský klient pro připojení k serverům protokolu Model Context Protocol a jejich testování, včetně nástrojů, zdrojů, promptů a odpovědí serveru.
Poslechněte si tento návodAI-vyprávěný · ~3 min
As of September 26, 2026, MCP has legacy revisions and a newer 2026-07-28 revision with a different versioning and transport lifecycle, so tests must target the protocol era and transport the server actually supports.
The Model Context Protocol (MCP), introduced by Anthropic, standardizes how LLM applications connect to external tools, data sources, and prompt templates. MCP Inspector is an interactive developer tool for testing and debugging MCP servers. Its current documentation describes Web, CLI, and TUI clients that can connect to local stdio processes or remote HTTP transports; a CLI mode can probe methods directly. The Streamable HTTP transport may return JSON or use server-sent events, while the older HTTP+SSE transport remains a legacy compatibility path. Once connected, Inspector lists everything the server declares - its tools, resources, and prompts - and lets a developer invoke any tool by filling in its parameters directly, without a language model in the loop at all. This matters because many integration bugs are not about model behavior; they are plain protocol or schema errors, such as a tool that returns the wrong content type, a resource URI that the server cannot actually resolve, or a parameter marked required in the schema but never checked in the handler. Inspector exposes server responses, tool results, metadata, and logs so developers can isolate protocol behavior before integrating a model. The MCP revision matters: 2025-era clients use an initialize/initialized exchange, while the 2026-07-28 revision removes that handshake and requires modern servers to support server/discover, which clients may call optionally. Check the configured revision as well as the transport when a connection or method behaves differently. A common misconception is that passing manual tests in Inspector guarantees a model will use the tools well in practice; it only confirms the server behaves correctly when called - whether a model chooses to call a tool, and with sensible arguments, still depends heavily on how clearly the tool and its parameters are described. Another misconception is that Inspector requires a live LLM API key; it does not, since it acts as its own MCP client.
A developer building a weather MCP server calls get_forecast in Inspector with a malformed date string and reads the exact JSON-RPC error returned, revealing a missing input validation check.
A team debugging why Claude never calls their search_docs tool loads the server in Inspector and notices the tool's description field is blank, explaining why the model skips it in favor of other tools.
Before shipping a new stdio-based MCP server, an engineer runs Inspector locally to confirm the server answers the initialize and tools/list requests correctly, catching a protocol version mismatch early.
A contributor to an open-source GitHub-issues MCP server uses Inspector's request history panel to compare the payload sent for list_issues against the parameter names documented in the README, finding a typo in a field name.
MCP Inspector and the protocol are changing together. The 2026-07-28 revision introduces a stateless protocol core and discovery-based capability lookup, while legacy revisions remain relevant for existing clients. Test the transport and revision combinations your deployment supports, and include CLI smoke tests for critical operations where appropriate. A successful Inspector call demonstrates one client request worked; it does not prove the server is secure, production-ready, or reliably selected by an AI host. Keep fixtures for both protocol eras if your clients still use both.
Newsletter zdarma
Tři ověřené příběhy ze světa AI každý všední den ráno, psané srozumitelnou angličtinou. Navždy zdarma, bez reklam.
Jeden e-mail každý všední den. Odhlásit se můžete jedním kliknutím. Vaši adresu nikdy neprodáváme ani nesdílíme.
Otestujte se
Okamžitá zpětná vazba ke každé odpovědi a sdílitelný certifikát s ověřitelným ID po absolvování kurzu.
Podpořte bezplatnou AI vzdělávání. AI Understanding je nezisková organizace 501(c)(3) — žádné reklamy, žádná platba za přístup, nikdy. Darujte
MCP Inspector is a developer client for connecting to and testing Model Context Protocol servers, including tools, resources, prompts, and server responses. As of September 26, 2026, MCP has legacy revisions and a newer 2026-07-28 revision with a different versioning and transport lifecycle, so tests must target the protocol era and transport the server actually supports.
Inspector is a standalone MCP client used to exercise a server's declared capabilities directly.
Inspector confirms the server behaves correctly when called, but model tool-choice depends on description quality, which is a separate concern.
For local servers, Inspector spawns the process and talks over stdio.
Legacy versions use the initialize exchange. In 2026-07-28, modern servers must implement server/discover, but clients may choose whether to call it before another request.
Vrácená chyba JSON-RPC je užitečný diagnostický důkaz, ale interpretujte ji spolu s požadavkem, metodou a chováním na straně serveru; Inspector automaticky neidentifikuje příčinu.
Učte se dál
Pro toto téma bylo vybráno více průvodců
DalšíDalší průvodce
Použití LLM Playgrounds k testování výzev
Technický