Ein schlanker MCP-Server, der es LLMs ermöglicht, andere MCP-Server zu inspizieren und zu testen. Dies ist eine in sich geschlossene Implementierung, die das MCP-SDK direkt verwendet, ohne auf externe CLIs zurückzugreifen.
Direkte SDK-Integration: Verwendet @modelcontextprotocol/sdk direkt für Server- und Client-Operationen
Alle Transporttypen: Unterstützt stdio-, SSE- und HTTP-Transporte (streamfähig)
Minimaler Fußabdruck: Einzige Abhängigkeit (@modelcontextprotocol/sdk)
Vollständige MCP-Inspektion: Tools auflisten, Tools aufrufen, Ressourcen auflisten, Ressourcen lesen, Prompts auflisten, Prompts abrufen
Sitzungsverwaltung: Persistente Verbindungen mit automatischer Garbage Collection
Ereignispufferung: Erfassung von Benachrichtigungen, Datenverkehr und Fehlern zur Fehlerbehebung
Related MCP server: Mock MCP Server
npm install
npm run buildFügen Sie dies zu Ihrer MCP-Konfiguration hinzu. Obwohl es leichte Unterschiede zwischen verschiedenen Implementierungen gibt, ist das allgemeine Format gleich:
{
"mcpServers": {
"mcp-inspector": {
"command": "node",
"args": ["/path/to/mcp-inspector-as-mcp-server/dist/server.js"]
}
}
}Tool | Beschreibung |
| Stellt eine persistente Verbindung zu einem MCP-Server her. Gibt eine |
| Schließt eine persistente Sitzung und gibt Ressourcen frei. |
| Listet alle aktiven Sitzungen mit ihrem Status und ihrer Leerlaufzeit auf. |
| Liest gepufferte Ereignisse (Benachrichtigungen, Datenverkehr, Fehler) aus einer Sitzung. |
| Injiziert eine menschliche Steuerungsnachricht in die Warteschlange einer Sitzung. |
Tool | Beschreibung |
| Listet alle von einem MCP-Server bereitgestellten Tools auf |
| Ruft ein Tool auf einem MCP-Server auf |
| Listet alle von einem MCP-Server bereitgestellten Ressourcen auf |
| Liest eine bestimmte Ressource |
| Listet Ressourcenvorlagen auf |
| Listet alle Prompts auf |
| Ruft einen bestimmten Prompt ab |
Alle Tools akzeptieren die folgenden Verbindungsparameter:
Für stdio-Transport (lokale Befehle):
command: Auszuführender Befehl (z. B. "node", "python")
args: Array von Argumenten (z. B. ["path/to/server.js"])
Für SSE/HTTP-Transport (entfernte Server):
url: Server-URL (z. B. "http://localhost:3000/sse")
headers: Optionales HTTP-Header-Objekt
Allgemein:
transport: Erzwingt den Transporttyp ("stdio", "sse" oder "http"). Wird automatisch erkannt, falls nicht angegeben.
session_id: (Optional) Verwendet eine bestehende persistente Sitzung anstelle einer kurzlebigen Verbindung.
Verwenden Sie für das Debugging von zustandsbehaftetem Serververhalten persistente Sitzungen:
1. insp_connect → returns session_id
2. insp_tools_list (with session_id) → uses persistent connection
3. insp_tools_call (with session_id) → state is preserved
4. insp_read_events (with session_id) → see notifications
5. insp_disconnect (with session_id) → cleanupSitzungen werden nach 30 Minuten Inaktivität automatisch geschlossen.
Der Inspector ermöglicht Human-in-the-Loop-Workflows, bei denen Sie LLM-gesteuerte MCP-Tests in Echtzeit beobachten und steuern können.
┌─────────────┐ MCP calls ┌─────────────────┐ forwards ┌─────────────┐
│ LLM Agent │ ◄────────────────► │ MCP Inspector │ ◄──────────────► │ Target MCP │
│ (Antigravity) │ (v2.0) │ │ Server │
└─────────────┘ └────────┬────────┘ └─────────────┘
│
Events logged to
session EventBuffer
│
┌───────────────────────┼───────────────────────┐
│ │ │
▼ ▼ ▼
insp_read_events HTTP :9847/api mcp-steer CLI
(LLM reads events) (external access) (human injection)Über LLM: Der Agent kann insp_read_events aufrufen, um zu sehen, was passiert:
{
"session_id": "sess_abc123",
"types": ["traffic_in", "traffic_out"],
"limit": 20
}Über HTTP: Fragen Sie die Steuerungs-API direkt ab:
curl http://127.0.0.1:9847/api/sessionsInjizieren Sie Anweisungen, die in der nächsten Tool-Antwort des LLMs erscheinen.
Über die CLI:
./bin/mcp-steer.mjs "Focus on testing the error handling paths"
./bin/mcp-steer.mjs --session sess_abc123 "Try calling with invalid params"Über HTTP:
curl -X POST http://127.0.0.1:9847/api/steer \
-H "Content-Type: application/json" \
-d '{"message": "Check the authentication flow next"}'Über das MCP-Tool:
{
"tool": "insp_inject_steering",
"arguments": {
"session_id": "sess_abc123",
"message": "Great progress! Now test edge cases."
}
}Typ | Beschreibung |
| Nachrichten, die AN den Zielserver gesendet wurden |
| Nachrichten, die VOM Zielserver empfangen wurden |
| MCP-Benachrichtigungen vom Zielserver |
| Fehler, die während der Kommunikation aufgetreten sind |
| Menschliche Steuerungsnachrichten, die in die Sitzung injiziert wurden |
LLM erstellt Sitzung: insp_connect → erhält sess_abc123
LLM startet Test: insp_tools_call mit session_id
Mensch beobachtet: curl http://127.0.0.1:9847/api/sessions
Mensch steuert: ./bin/mcp-steer.mjs "Also test the batch endpoint"
LLM empfängt Steuerung: Die nächste Tool-Antwort enthält ⚡ STEERING from human: ...
LLM passt sich an: Berücksichtigt die menschliche Anleitung
Tools von einem lokalen MCP-Server auflisten (kurzlebig):
{
"command": "node",
"args": ["/path/to/some-mcp-server/dist/server.js"]
}Eine persistente Sitzung erstellen:
{
"command": "node",
"args": ["/path/to/some-mcp-server/dist/server.js"]
}
// Returns: { "session_id": "sess_abc123", "server_info": {...} }Ein Tool über eine Sitzung aufrufen:
{
"session_id": "sess_abc123",
"tool_name": "search",
"tool_args": {"query": "hello"}
}├── src/
│ ├── server.ts # MCP server exposing inspector tools
│ ├── client.ts # Client wrapper (hybrid stateless/session mode)
│ ├── transport.ts # Transport factory (stdio, SSE, HTTP) + TracingWrapper
│ ├── session.ts # SessionRegistry with GC (30-min TTL)
│ └── events.ts # EventBuffer (ring buffer for notifications)
├── bin/
│ └── mcp-steer.mjs # CLI tool for human steering
├── tests/ # Integration test scripts (run with npx tsx)
└── vitest.config.ts # Unit test + coverage configDer ursprüngliche MCP Inspector ist eine Kombination aus Web-UI und CLI, die über mehrere Projekte verteilt ist. Dies konsolidiert die Kernfunktionalität in einem einzigen, schlanken MCP-Server, den ein LLM verwenden kann, um:
MCP-Server iterativ zu entwickeln und zu debuggen
MCP-Server-Funktionalität zu testen, ohne die Konversation zu verlassen
Zu erkunden, welche Tools/Ressourcen/Prompts ein MCP-Server bereitstellt
Zustandsbehaftetes Verhalten mit persistenten Sitzungen zu debuggen
npm install # install dependencies
npm run build # compile TypeScript
npm run dev # watch mode
npm test # run unit tests
npm run test:cov # run tests with coverage
npm run lint # lint source files
npm run format # auto-format with Prettier
npm run typecheck # type-check without emittingMenschliche Steuerung (insp_inject_steering) für Human-in-the-Loop-Workflows hinzugefügt
HTTP-API auf Port 9847 für externe Steuerung/Beobachtbarkeit hinzugefügt
mcp-steer.mjs CLI-Tool für einfache menschliche Interaktion hinzugefügt
Kritischen Fehler in TracingTransportWrapper behoben, bei dem das Timing der Handler-Erfassung zu Nachrichtenverlust führte
Sitzungsverwaltung hinzugefügt (insp_connect, insp_disconnect, insp_list_sessions)
Ereignispufferung hinzugefügt (insp_read_events)
Alle Inspektions-Tools unterstützen jetzt eine optionale session_id für persistente Verbindungen
Automatische Garbage Collection hinzugefügt (30-Minuten-TTL für inaktive Sitzungen)
Abwärtskompatibel: session_id weglassen für das ursprüngliche kurzlebige Verhalten
Erstveröffentlichung mit kurzlebigen Verbindungen
MIT
Get a specific prompt from an MCP server.
| Name | Required | Description | Default |
|---|---|---|---|
| command | No | Command to run the MCP server (e.g., 'node', 'python') | |
| args | No | Arguments to pass to the command (e.g., ['build/index.js']) | |
| url | No | URL for SSE/HTTP transport (alternative to command) | |
| transport | No | Transport type (auto-detected if not specified) | |
| headers | No | HTTP headers for SSE/HTTP transport | |
| prompt_name | Yes | Name of the prompt to get | |
| prompt_args | No | Arguments to pass to the prompt |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries full burden for behavioral disclosure but offers minimal information. It states the tool retrieves a prompt but doesn't describe what happens if the prompt doesn't exist, whether authentication is required, if there are rate limits, what format the prompt returns in, or whether this is a read-only operation. The description is too basic for a tool with 7 parameters and server interaction.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single, clear sentence that gets straight to the point with zero wasted words. It's appropriately sized for a retrieval operation and front-loads the essential information. Every word earns its place in communicating the core functionality.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a tool with 7 parameters, server communication, and no output schema, the description is insufficiently complete. It doesn't explain what constitutes a 'prompt' in this system, what the return format looks like, error handling, or authentication requirements. The combination of complex parameters and no annotations means the description should provide more contextual information about the operation.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, so the schema already documents all 7 parameters thoroughly. The description adds no additional parameter information beyond what's in the schema - it doesn't explain relationships between parameters (like command/args vs url/transport), provide examples of prompt_name formats, or clarify when prompt_args are needed. This meets the baseline for high schema coverage.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the action ('Get') and resource ('a specific prompt from an MCP server'), making the purpose immediately understandable. It distinguishes from sibling tools like 'insp_prompts_list' by specifying retrieval of a single prompt rather than listing multiple. However, it doesn't explicitly mention what 'prompt' refers to in this context (e.g., AI prompt templates, system prompts).
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides no guidance on when to use this tool versus alternatives. It doesn't mention when to choose this over 'insp_prompts_list' (for listing all prompts) or 'insp_tools_call' (which might handle different operations). There's no discussion of prerequisites, error conditions, or typical use cases for prompt retrieval.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
List all prompts exposed by an MCP server.
| Name | Required | Description | Default |
|---|---|---|---|
| command | No | Command to run the MCP server (e.g., 'node', 'python') | |
| args | No | Arguments to pass to the command (e.g., ['build/index.js']) | |
| url | No | URL for SSE/HTTP transport (alternative to command) | |
| transport | No | Transport type (auto-detected if not specified) | |
| headers | No | HTTP headers for SSE/HTTP transport |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries full burden for behavioral disclosure. It states it's a list operation, implying read-only behavior, but doesn't mention any side effects, permissions required, rate limits, or what the output format looks like. For a tool that interacts with external servers, this lack of operational context is a significant gap.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single, clear sentence that efficiently conveys the core purpose without any fluff. It's front-loaded with the main action and resource, making it easy to parse. Every word earns its place in defining what the tool does.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the complexity of connecting to external servers via multiple transport methods and the lack of annotations and output schema, the description is insufficient. It doesn't explain what 'prompts' are in this context, how results are returned, or any error conditions. For a tool with 5 parameters and no structured safety hints, more operational detail is needed.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, so the schema fully documents all 5 parameters. The description adds no parameter-specific information beyond implying the tool connects to an MCP server. This meets the baseline of 3 where the schema does the heavy lifting, but the description doesn't compensate with additional context like default behaviors or parameter interactions.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the action ('List all prompts') and the target resource ('exposed by an MCP server'), making the purpose immediately understandable. It distinguishes from siblings like insp_tools_list by specifying 'prompts' rather than 'tools', but doesn't explicitly contrast with insp_prompts_get, which would fetch a specific prompt rather than list all.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides no guidance on when to use this tool versus alternatives like insp_prompts_get or insp_tools_list. It mentions the scope ('all prompts') but offers no context about prerequisites, typical use cases, or limitations that would help an agent decide between this and sibling tools.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
List all resources exposed by an MCP server.
| Name | Required | Description | Default |
|---|---|---|---|
| command | No | Command to run the MCP server (e.g., 'node', 'python') | |
| args | No | Arguments to pass to the command (e.g., ['build/index.js']) | |
| url | No | URL for SSE/HTTP transport (alternative to command) | |
| transport | No | Transport type (auto-detected if not specified) | |
| headers | No | HTTP headers for SSE/HTTP transport |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries the full burden of behavioral disclosure. It states it 'lists' resources, implying a read-only operation, but doesn't cover aspects like whether it requires authentication, how it handles errors, if it's rate-limited, or what the output format looks like (e.g., JSON list). This leaves significant gaps for an agent to understand its behavior.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single, clear sentence that front-loads the core purpose without unnecessary words. It efficiently conveys the essential information, making it easy for an agent to parse quickly.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the complexity (5 parameters, no annotations, no output schema), the description is insufficient. It doesn't explain what 'resources' entail in this context, how results are returned, or any behavioral traits like error handling. For a tool that likely inspects server capabilities, more context is needed to guide effective use.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The input schema has 100% description coverage, so parameters are well-documented in the schema itself. The description adds no additional meaning about parameters beyond implying the tool interacts with an MCP server, which is already inferred from the schema's command/args/url fields. This meets the baseline for high schema coverage.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the action ('List all resources') and the target ('exposed by an MCP server'), providing a specific verb+resource combination. However, it doesn't explicitly differentiate from sibling tools like 'insp_resources_read' or 'insp_resources_templates', which likely have different purposes related to resources.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides no guidance on when to use this tool versus alternatives. With siblings like 'insp_resources_read' (likely for reading a specific resource) and 'insp_resources_templates' (likely for templates), there's no indication of context, prerequisites, or exclusions for this list operation.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Read a specific resource from an MCP server.
| Name | Required | Description | Default |
|---|---|---|---|
| command | No | Command to run the MCP server (e.g., 'node', 'python') | |
| args | No | Arguments to pass to the command (e.g., ['build/index.js']) | |
| url | No | URL for SSE/HTTP transport (alternative to command) | |
| transport | No | Transport type (auto-detected if not specified) | |
| headers | No | HTTP headers for SSE/HTTP transport | |
| uri | Yes | URI of the resource to read |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries the full burden of behavioral disclosure. It states 'Read a specific resource,' implying a read-only operation, but doesn't cover critical aspects like authentication needs, rate limits, error handling, or what the output looks like (e.g., raw data, structured format). For a tool with 6 parameters and no output schema, this is a significant gap in transparency.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single, clear sentence that directly states the tool's purpose. It's front-loaded with the core action ('Read a specific resource') and avoids unnecessary details. Every word earns its place, making it highly concise and well-structured for quick understanding.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the tool's complexity (6 parameters, nested objects, no output schema) and lack of annotations, the description is incomplete. It doesn't explain the resource type, how parameters like 'transport' affect behavior, or what the read operation returns. For a tool that likely involves server interaction and resource retrieval, more context is needed to guide effective use.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, meaning all parameters are documented in the input schema. The description adds no additional meaning beyond the schema, such as explaining how parameters interact (e.g., 'command' vs. 'url' for transport) or providing examples. With high schema coverage, the baseline score of 3 is appropriate, as the description doesn't compensate but also doesn't detract.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description states the tool 'Read a specific resource from an MCP server,' which clearly indicates a read operation on a resource. However, it doesn't specify what type of resource (e.g., file, data object) or differentiate from sibling tools like 'insp_resources_list' (which likely lists resources) or 'insp_resources_templates' (which might handle templates). The purpose is clear but lacks sibling differentiation.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides no guidance on when to use this tool versus alternatives. It doesn't mention prerequisites (e.g., server setup), exclusions (e.g., not for writing), or compare to siblings like 'insp_resources_list' for listing resources. Without such context, an agent might struggle to select the correct tool in a given scenario.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
List resource templates exposed by an MCP server.
| Name | Required | Description | Default |
|---|---|---|---|
| command | No | Command to run the MCP server (e.g., 'node', 'python') | |
| args | No | Arguments to pass to the command (e.g., ['build/index.js']) | |
| url | No | URL for SSE/HTTP transport (alternative to command) | |
| transport | No | Transport type (auto-detected if not specified) | |
| headers | No | HTTP headers for SSE/HTTP transport |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries the full burden of behavioral disclosure. It states the tool lists resource templates but doesn't describe what 'exposed by an MCP server' entails, such as whether this requires server connectivity, authentication, or specific permissions. For a tool with 5 parameters and no annotation coverage, this is a significant gap in transparency.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single, efficient sentence that directly states the tool's purpose without unnecessary words. It's front-loaded and wastes no space, making it easy to understand at a glance while being appropriately sized for its function.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the complexity (5 parameters, no annotations, no output schema), the description is minimal but covers the basic purpose. It lacks details on behavioral aspects like server interaction requirements or output format, which are important for a tool that likely involves external communication. However, the high schema coverage mitigates some gaps, making it adequate but with clear room for improvement.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The input schema has 100% description coverage, providing clear details for all 5 parameters (e.g., command, args, url, transport, headers). The description doesn't add any parameter-specific information beyond what's in the schema, such as examples or usage context. With high schema coverage, the baseline score of 3 is appropriate as the schema handles the heavy lifting.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the action ('List') and target ('resource templates exposed by an MCP server'), providing a specific verb+resource combination. However, it doesn't explicitly distinguish this from sibling tools like 'insp_resources_list' or 'insp_resources_read', which likely handle different aspects of resources, leaving some ambiguity about differentiation.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides no guidance on when to use this tool versus alternatives. It doesn't mention any prerequisites, context for usage, or comparisons to sibling tools such as 'insp_resources_list', which might handle actual resources rather than templates. This lack of guidance could lead to confusion in tool selection.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Call a tool on an MCP server. Connects, calls the tool, and disconnects.
| Name | Required | Description | Default |
|---|---|---|---|
| command | No | Command to run the MCP server (e.g., 'node', 'python') | |
| args | No | Arguments to pass to the command (e.g., ['build/index.js']) | |
| url | No | URL for SSE/HTTP transport (alternative to command) | |
| transport | No | Transport type (auto-detected if not specified) | |
| headers | No | HTTP headers for SSE/HTTP transport | |
| tool_name | Yes | Name of the tool to call | |
| tool_args | No | Arguments to pass to the tool (key=value pairs) |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries the full burden of behavioral disclosure. It mentions connecting, calling, and disconnecting, which implies network/process operations, but doesn't disclose critical traits like error handling, timeouts, authentication needs, rate limits, or what happens if the server is unavailable. For a tool that interacts with external servers, this lack of behavioral context is a significant gap.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is extremely concise (one sentence) and front-loaded with the core purpose. Every word earns its place by summarizing the tool's lifecycle (connect, call, disconnect). There's no redundancy or fluff, making it efficient for quick understanding.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the tool's complexity (7 parameters, no annotations, no output schema), the description is incomplete. It doesn't address what the tool returns, error conditions, or how to interpret results from the called tool. For a tool that dynamically invokes other tools on a server, more context about output format, success/failure states, and integration patterns is needed to be fully helpful.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, so the schema already documents all 7 parameters thoroughly. The description adds no additional meaning beyond what's in the schema (e.g., it doesn't explain parameter interactions or provide examples). With high schema coverage, the baseline score of 3 is appropriate, as the description doesn't compensate but also doesn't detract from the well-documented schema.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool's purpose: 'Call a tool on an MCP server' with specific verbs (connects, calls, disconnects). It distinguishes from siblings like insp_tools_list (which lists tools) but doesn't explicitly contrast with other tools that might also involve calling operations. The purpose is well-defined but could be more specific about what distinguishes it from potential alternatives.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides no guidance on when to use this tool versus alternatives. It doesn't mention siblings like insp_tools_list (which might be used to discover tools before calling) or other tools that might handle MCP server interactions differently. There's no context about prerequisites, error conditions, or typical use cases, leaving the agent with minimal usage direction.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
List all tools exposed by an MCP server. Connects, lists tools, and disconnects.
| Name | Required | Description | Default |
|---|---|---|---|
| command | No | Command to run the MCP server (e.g., 'node', 'python') | |
| args | No | Arguments to pass to the command (e.g., ['build/index.js']) | |
| url | No | URL for SSE/HTTP transport (alternative to command) | |
| transport | No | Transport type (auto-detected if not specified) | |
| headers | No | HTTP headers for SSE/HTTP transport |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries the full burden of behavioral disclosure. It mentions the connection and disconnection process, which is helpful, but lacks critical details such as whether this is a read-only operation, potential side effects (e.g., server state changes), error handling, or performance considerations (e.g., timeouts). For a tool that interacts with external servers, this is a significant gap.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is extremely concise—just one sentence with three clauses—and front-loaded with the core purpose. Every word earns its place by conveying essential information about the tool's function and operational flow without any redundancy or fluff.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the complexity of connecting to and querying an MCP server, the description is incomplete. It lacks details on output format (no output schema is provided), error conditions, authentication needs, or rate limits. While the schema covers parameters well, the overall context for safe and effective use is insufficient, especially for a tool with external dependencies.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The schema description coverage is 100%, meaning all parameters are documented in the input schema. The description adds no additional parameter semantics beyond what's in the schema (e.g., it doesn't explain parameter interactions or provide examples). This meets the baseline score of 3 for high schema coverage, but doesn't compensate with extra value.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the action ('List all tools') and resource ('exposed by an MCP server'), providing a specific verb+resource combination. It also mentions the operational flow ('Connects, lists tools, and disconnects'), which adds useful context. However, it doesn't explicitly differentiate this tool from its sibling 'insp_tools_call', which appears to be for invoking tools rather than listing them.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides no guidance on when to use this tool versus alternatives like 'insp_tools_call' or other sibling tools. It mentions the operational steps but doesn't specify prerequisites, use cases, or exclusions. This leaves the agent without clear direction on tool selection in context.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
v1.0.0insp_prompts_getinsp_prompts_listinsp_resources_listinsp_resources_readinsp_resources_templatesinsp_tools_callinsp_tools_listScored across 7 tools
Every tool has a clearly distinct purpose targeting different MCP server components: prompts (get/list), resources (list/read/templates), and tools (list/call). There is no overlap or ambiguity in functionality, making it easy for an agent to select the correct tool.
All tools follow a consistent 'insp_[component]_[action]' pattern with snake_case, using clear verbs like get, list, read, call, and templates. This predictability enhances usability and reduces confusion.
With 7 tools, the server is well-scoped for inspecting MCP servers, covering prompts, resources, and tools comprehensively. Each tool earns its place without being excessive or insufficient for the domain.
The tool set provides complete coverage for inspecting MCP servers, including listing and accessing prompts, resources (with templates), and tools (with calling capability). There are no obvious gaps, ensuring agents can perform all necessary inspection tasks.
MCP server for your apps' tools and custom tools, plus hosted AI agents and approval-gated workflows
The Remote MCP server acts as a standardized bridge between LLM applications (like Claude, ChatGPT, and Cursor) and external services, enabling AI agents to access external tools and resources. Its primary capability is providing a centralized search tool to discover other MCP servers and their respective tools. Unlike local implementations, it runs remotely with OAuth authentication and permission controls for security.
Security scanner for MCP servers. Detect vulnerabilities, prompt injection, and tool poisoning.
Remote MCP server for supportsheep: run AI interviews and manage support content for your blog.