Skip to main content
Glama

Audit an MCP server for tool poisoning

audit_mcp_server
Read-only

Check an MCP server for tool poisoning before connecting: hidden instructions, sensitive-file access, exfiltration parameters, cross-tool shadowing, rug pulls. For skills or AGENTS.md use audit_agent_skill. Give url (tools are listed read-only, never called) or the pasted tools, not both. Returns level, findings (type, severity, location, snippet) and per-tool fingerprints to send back as pinned. Costs US$0.01 in USDC on Base or Solana (x402); servers that cannot be listed are not charged.

Tool call template

This tool declares no required parameters. Add optional parameters from the input schema as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {},
    "name": "audit_mcp_server"
  }
}

Input Schema

NameRequiredDescriptionDefault
urlNoStreamable HTTP endpoint of a public MCP server. Omit when pasting tools
toolsNoInstead of url: the tools array of a tools/list result, pasted as is
pinnedNoOptional map of tool name to the fingerprint returned by a previous audit; any change is reported as tool-changed (high)
instructionsNoOnly with tools: the server's instructions from its initialize result

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed4 schema fields changed
    • changedInput schema / properties / instructions / description
      Previous value: -"Server instructions, if pasting"New value: +"Only with tools: the server's instructions from its initialize result"
    • changedInput schema / properties / pinned / description
      Previous value: -"Tool name to fingerprint from a previous audit"New value: +"Optional map of tool name to the fingerprint returned by a previous audit; any change is reported as tool-changed (high)"
    • changedInput schema / properties / tools / description
      Previous value: -"Or the tools/list result, pasted"New value: +"Instead of url: the tools array of a tools/list result, pasted as is"
    • changedInput schema / properties / url / description
      Previous value: -"Streamable HTTP endpoint of the MCP server"New value: +"Streamable HTTP endpoint of a public MCP server. Omit when pasting tools"
  2. First observed

TDQS

A4.8/5.0
Behavior5/5

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

Annotations already declare readOnlyHint=true and openWorldHint=true, and the description reinforces rather than contradicts this with the concrete guarantee that tools from the target server are "listed read-only, never called." It goes well beyond the annotations by disclosing the payment mechanism and price (US$0.01 USDC on Base or Solana via x402), the non-billing failure case, and the return shape including fingerprints for re-audit.

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

Conciseness4/5

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

Purpose is front-loaded and every sentence carries load-bearing information (threats, sibling routing, input exclusivity, return shape, cost). It is tightly packed with parentheticals and clauses, which costs a little readability, but there is no filler to cut.

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 paid, open-world audit tool with no output schema, the description covers all the decision-critical facts: what is inspected, what is and isn't touched on the target, how inputs are supplied, what comes back (level, findings, fingerprints), and the billing/auth prerequisite. Nothing an agent needs before invoking it is missing.

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

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, so baseline is 3, but the description adds real meaning: it frames url vs tools as mutually exclusive inputs and explains the round-trip purpose of pinned (send back previously returned fingerprints so drift is flagged). It does not clarify the instructions parameter, which the schema alone documents.

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?

States a specific verb and resource ("Check an MCP server for tool poisoning") and enumerates the concrete threat classes it detects: hidden instructions, sensitive-file access, exfiltration parameters, cross-tool shadowing, rug pulls. It explicitly names the sibling audit_agent_skill as the tool for a different resource (skills or AGENTS.md), so an agent can route without opening either schema.

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?

Gives the temporal condition ("before connecting"), the exact alternative and when to pick it (audit_agent_skill for skills/AGENTS.md), and an explicit mutual-exclusion rule ("Give url ... or the pasted tools, not both"). It also states when no charge applies ("servers that cannot be listed are not charged"), which affects whether an agent should retry or escalate.

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

Try in Browser

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources