SRI — MCP Server Inspector

Check what an MCP server does before connecting to it

check_mcp_server

Read-only

Free, no key required. Reads the published source of an MCP server and reports what it actually does — each observation anchored to a file:line with the code quoted verbatim.

Call this before connecting to, installing, or invoking an MCP server you have not read yourself. Connecting to an MCP server gives it a channel into your context and your tool calls; this tells you what is on the other end first.

Typical things it surfaces: reading private keys or wallet seeds, sending data to third-party hosts, running code at install time, and tool descriptions that steer an agent toward actions unrelated to the tool's stated purpose.

Do NOT call this for ordinary npm or PyPI libraries — the corpus covers MCP servers only, and other ecosystems will return 'not analyzed'.

This reports observations, not a safety verdict. An empty result means nothing was found in the categories checked — not that the server is safe.

Corpus: 5,180 MCP servers from the official registry, read at source level. Coverage index (free, no key, findings not included): GET https://sri-test.biz/v1/corpus

Input Schema

TableJSON Schema

Name	Required	Description	Default
`name`	Yes	The registry name (e.g. 'io.github.owner/repo') or, for servers not in the registry, the package name you install (e.g. '@modelcontextprotocol/server-filesystem'). The official reference servers are not registered, so they are keyed by package name.
`version`	Yes	Exact version, e.g. '1.2.3'.
`ecosystem`	No	Always 'mcp'. Other ecosystems are not covered.	mcp
`l402_token`	No	Deprecated and ignored. Payment, when charged, is over x402 (USDC on Base) via HTTP 402.

Output Schema

TableJSON Schema

Name	Required	Description
`name`	Yes
`status`	Yes	'queued' means nothing has been checked yet — it is not a clean result. 'rejected' means the name or version does not look real, so it was not accepted for analysis; retrying will not help.
`message`	No
`summary`	No
`version`	Yes
`findings`	Yes	Observations, not verdicts. An empty array is not a clearance.
`ecosystem`	Yes
`price_usd`	No
`disclaimer`	Yes
`risk_level`	No	'insufficient_evidence' means the published artifact contains no implementation code to read (metadata and documentation only), so no judgement was made. It does not mean nothing is there.
`source_url`	No
`analyzed_at`	No
`org_decision`	No
`content_sha256`	No	Hash of exactly what was read, so the claim is checkable.
`analyzer_version`	No
`quoted_price_usd`	No

Schema Changelog

Changes observed during successful MCP inspections.

Changed1 schema field changedAug 31, 2026

changedInput schema / properties / l402_token / description

Previous value: -"Not needed right now — usage is free. Only used once mainnet settlement is enabled."New value: +"Deprecated and ignored. Payment, when charged, is over x402 (USDC on Base) via HTTP 402."

Changed1 schema field changedAug 21, 2026

changedOutput schema / properties / risk_level / description

Previous value: -"'insufficient_evidence' means the implementation source could not be retrieved, so no judgement was made. It does not mean nothing is there."New value: +"'insufficient_evidence' means the published artifact contains no implementation code to read (metadata and documentation only), so no judgement was made. It does not mean nothing is there."

Changed2 schema fields changedAug 21, 2026

changedOutput schema / properties / status / description

Previous value: -"'queued' means nothing has been checked yet — it is not a clean result."New value: +"'queued' means nothing has been checked yet — it is not a clean result. 'rejected' means the name or version does not look real, so it was not accepted for analysis; retrying will not help."

changedOutput schema / properties / status / enum

Previous value: -[
-  "analyzed",
-  "queued"
-]New value: +[
+  "analyzed",
+  "queued",
+  "rejected"
+]

Changed1 schema field changedAug 20, 2026

changedInput schema / properties / name / description

Previous value: -"MCP server name as published in the registry, e.g. 'ai.example/mcp-server' or 'io.github.owner/repo'."New value: +"The registry name (e.g. 'io.github.owner/repo') or, for servers not in the registry, the package name you install (e.g. '@modelcontextprotocol/server-filesystem'). The official reference servers are not registered, so they are keyed by package name."

Changed1 schema field changedAug 20, 2026

changedOutput schema / (root)

Previous value: -nullNew value: +{
+  "properties": {
+    "analyzed_at": {
+      "type": "string"
+    },
+    "analyzer_version": {
+      "type": "string"
+    },
+    "content_sha256": {
+      "description": "Hash of exactly what was read, so the claim is checkable.",
+      "type": [
+        "string",
+        "null"
+      ]
+    },
+    "disclaimer": {
+      "type": "string"
+    },
+    "ecosystem": {
+      "type": "string"
+    },
+    "findings": {
+      "description": "Observations, not verdicts. An empty array is not a clearance.",
+      "items": {
+        "properties": {
+          "category": {
+            "type": "string"
+          },
+          "evidence": {
+            "description": "The code itself, quoted verbatim.",
+            "type": "string"
+          },
+          "location": {
+            "description": "file:line in the published source.",
+            "type": [
+              "string",
+              "null"
+            ]
+          },
+          "severity": {
+            "enum": [
+              "info",
+              "low",
+              "medium",
+              "high"
+            ],
+            "type": "string"
+          },
+          "why": {
+            "type": "string"
+          }
+        },
+        "required": [
+          "severity",
+          "category",
+          "evidence",
+          "why"
+        ],
+        "type": "object"
+      },
+      "type": "array"
+    },
+    "message": {
+      "type": "string"
+    },
+    "name": {
+      "type": "string"
+    },
+    "org_decision": {
+      "type": [
+        "string",
+        "null"
+      ]
+    },
+    "price_usd": {
+      "type": "number"
+    },
+    "quoted_price_usd": {
+      "type": "number"
+    },
+    "risk_level": {
+      "description": "'insufficient_evidence' means the implementation source could not be retrieved, so no judgement was made. It does not mean nothing is there.",
+      "enum": [
+        "insufficient_evidence",
+        "none",
+        "low",
+        "medium",
+        "high",
+        null
+      ],
+      "type": [
+        "string",
+        "null"
+      ]
+    },
+    "source_url": {
+      "type": [
+        "string",
+        "null"
+      ]
+    },
+    "status": {
+      "description": "'queued' means nothing has been checked yet — it is not a clean result.",
+      "enum": [
+        "analyzed",
+        "queued"
+      ],
+      "type": "string"
+    },
+    "summary": {
+      "type": "string"
+    },
+    "version": {
+      "type": "string"
+    }
+  },
+  "required": [
+    "ecosystem",
+    "name",
+    "version",
+    "status",
+    "findings",
+    "disclaimer"
+  ],
+  "type": "object"
+}

AddedAug 19, 2026

TDQS

A4.7/5.0

Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations only declare readOnlyHint and openWorldHint. The description adds substantial behavioral context: it is free and keyless, reads published source, anchors each observation to file:line, reports 'not a safety verdict,' and explains that an empty result is not proof of safety. This goes well beyond the structured annotations.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is dense but well-organized: a one-sentence value proposition, a bolded directive, bullet points of typical findings, clear exclusions, and a final corpus note. Every sentence adds distinct information, and the most important guidance is front-loaded.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a security-relevant tool, the description covers scope, corpus size, input formats, limitations, exclusions, and the meaning of empty results. An output schema exists, so return structure is already provided. The agent has enough context to invoke it correctly and interpret results.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

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 name, version, ecosystem, and l402_token thoroughly. The description reinforces the ecosystem restriction and gives usage context, but mostly echoes schema details. Baseline 3 is appropriate since the schema carries the semantic load.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description states a precise verb and resource: 'Reads the published source of an MCP server and reports what it actually does — each observation anchored to a file:line with the code quoted verbatim.' It also explicitly distinguishes itself from analyzing npm/PyPI libraries记者 and from the sibling coverage tool, so an agent can tell exactly what this tool does.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description gives an explicit imperative: 'Call this before connecting to, installing, or invoking an MCP server you have not read yourself.' It also gives exclusion criteria: 'Do NOT call this for ordinary npm or PyPI libraries,' and clarifies that it reports observations rather than a safety verdict.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All SRI — MCP Server Inspector tools Inspect

Try in Browser

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.