A read-only checker for Model Context Protocol (MCP) servers, which let AI applications use tools and data from other programs. It reviews source code against the 2026-07-28 MCP specification, tool design, and security practices without running the server.

Use it to inspect an MCP server's SDK, transport, lifecycle, extensions, OAuth/OIDC authorization, tools, and security controls, then produce findings with source-file evidence.

It finds protocol mismatches, unsafe designs, and authorization problems before deployment, without making requests or exposing secrets.

mcp-audit — Command for Claude Code by sigistry/marketplace

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Clone the repo

git clone --depth 1 https://github.com/sigistry/marketplace

Made for: Claude Code.

Or install mcp-server-smith, the plugin that ships this one along with the rest of its 4 skills, 3 commands, 2 agents.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/commands/sigistry/marketplace/mcp-audit/github.svg)](https://agentmods.dev/commands/sigistry/marketplace/mcp-audit)

Your own site

<a href="https://agentmods.dev/commands/sigistry/marketplace/mcp-audit"><img src="https://agentmods.dev/badge/commands/sigistry/marketplace/mcp-audit/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/commands/sigistry/marketplace/mcp-audit"><img src="https://agentmods.dev/badge/commands/sigistry/marketplace/mcp-audit.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 24 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 1,298 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 1 finding. A grade says what 26 rules found in the file — not that it is safe.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5.1	$0.00024	$0.01298
Opus 5.5	$0.00010	$0.00519
Sonnet 5.5	$0.00005	$0.00260
Haiku 4.5	$0.00002	$0.00130

Measured 1mo ago against content hash 587a9d94d54e, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

mcp-audit scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Runs shell commandslowCapability

Expected in a hook, worth knowing in a rule or an instructions file.

Apply `mcp-server-security` (`references/threat-model.md`): input validation on every tool argument; no shell/SQL/path injection from arguments (command construction, `child_process`, `eval`, unparameterized queries, pat

plugins/mcp-server-smith/commands/mcp-audit.md · 53 lines

How it starts

The opening of the file, as written. The whole thing — 53 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Audit the MCP server rooted at $ARGUMENTS (a directory, an entry file, or the current repo if omitted) for compliance with the MCP 2026-07-28 specification, tool-design quality, and security. This is a static, read-only review: read the source, never start the server, never send a request, never read secrets. Load the mcp-2026-07-28 skill (transport + lifecycle + extensions), the mcp-authorization skill (OAuth/OIDC), the mcp-tool-design skill, and the mcp-server-security skill, then dispatch the spec-compliance-auditor agent to produce the findings.

Process

Step 1: Detect the SDK, transport, and spec era

Glob for the entry point and manifest (package.json, pyproject.toml, server.json, Dockerfile). Identify the SDK and language:

Ecosystem	Import / symbol to look for	Server class
TypeScript	`@modelcontextprotocol/sdk`	`McpServer` / `Server`, `StreamableHTTPServerTransport`, `StdioServerTransport`
Python	`mcp` / `fastmcp`	`FastMCP`, `Server`, `streamable_http_app`, `stdio_server`
Go	`mcp-go` / `modelcontextprotocol/go-sdk`	`server.NewMCPServer`, `StreamableHTTP`

Determine the transport (stdio, Streamable HTTP, or the removed HTTP+SSE) and the declared protocol version (the protocolVersion in the initialize result, or the SDK version pinned in the manifest). Flag a server still on HTTP+SSE or advertising a pre-2026-07-28 version as the primary migration finding.

Step 2: Score compliance against 2026-07-28

Check, with file:line evidence, each axis from mcp-2026-07-28:

Stateless core. No per-connection server memory required across requests; each request is self-contained; safe to run on serverless/edge behind multiple instances. Flag reliance on in-process session state that a load balancer would shard incorrectly (see references/transport-and-lifecycle.md).
Lifecycle. Correct initialize handshake, capability negotiation, and version reporting; no assumption of a long-lived bidirectional channel.
Versioned extensions. If the server ships MCP Apps (interactive UI) or Tasks (long-running work), confirm they are declared under the versioned extensions framework, not bolted onto core methods.
Capabilities. Declared capabilities (tools, resources, prompts, logging, completions) actually match what the handlers implement.

Read the full file on GitHub · 53 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

1mo ago First seen · 53 lines · 24 tokens per session scan A 587a9d94d54e

Subscribe to this mod's changes

mcp-audit is a command published in the GitHub repository sigistry/marketplace (3 stars, last pushed 4d ago), licensed MIT. It adds 24 tokens to every session and 1,298 once invoked, about $0.0001 per session on Opus 5.5. A static security scan graded it A with 1 finding (runs shell commands). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.

Building agents Claude Code Security agent-skills claude claude-code-marketplace claude-code-plugin marketplace

mcp-audit

Getting it into your agent

Wrote this? Show the measurements

What it costs to keep this loaded

Grade A, and why

Runs shell commandslowCapability

How it starts

Process

Step 1: Detect the SDK, transport, and spec era

Step 2: Score compliance against 2026-07-28

What this file has done since we first saw it

Other commands, from other repositories

slides

typescript-dev

dotnet-structure

review

audit

plansession