広告掲載
MCP.so
ログイン
M

MCP Guardian

@rudraneel93

リポジトリ

基本情報

カテゴリ

開発者ツール

トランスポート

stdio

作者

rudraneel93

投稿者

Rudraneel Das

MCP Guardian について

> MCP Guardian is a runtime security, cost governance, and health monitoring proxy for MCP infrastructure. Three-layer detection engine (regex → schema → LLM), YAML policy engine with hot-reload, 50+ secret patterns + Shannon entropy, AST command validation, JWT algorithm pinning

MCP Guardian の設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

{
  "mcpServers": {
    "mcp-guardian": {
      "command": "npx",
      "args": [
        "-y",
        "@mcp-guardian/server"
      ]
    }
  }
}

MCP Guardian のツール

60

Scan MCP server configurations for security vulnerabilities (CVEs, auth, typo-squatting, secrets)

Audit token usage and estimate costs per MCP server

Check health, latency, and reliability of MCP servers

Generate a complete security, cost, and health report for all MCP servers

Start observing AI agent tool calls to learn usage patterns for policy generation

Stop the current observation window and finalize collected data

Generate a minimal-privilege YAML policy based on observed tool call patterns

Compare observed behavior against current policy and suggest additions/removals

Get current behavior observation status and summary

Scan tool call arguments for prompt injection payloads targeting downstream AI agents

Get prompt injection detection statistics

Generate threat forecast for all configured MCP servers with 30/90/365-day projections

Detailed threat forecast for a specific server with risk factors and preemptive hardening recommendations

Get suggested preemptive policy changes based on threat forecasts

Full supply chain integrity verification with signed attestation for MCP server packages

Current trust graph state for all MCP server packages

Export Software Bill of Materials for MCP server packages

Compare current MCP server behavior against a known-good baseline to detect anomalies

Capture current server state as a known-good behavioral baseline

Revert to a previous known-good configuration snapshot

List all detected drift events

Generate auditor-ready compliance evidence bundle for a framework

Identify missing compliance controls and recommend policies

Get current compliance posture score across all frameworks

List all supported compliance frameworks

Run a full autonomous red team assessment with attack generation and policy testing

Configure periodic autonomous red team assessments

Get latest red team assessment results and recommendations

A/B test a proposed policy change against historical attack corpus

Submit an anonymized threat signature to the cross-deployment intelligence mesh

Get mesh connectivity, contribution stats, and known threat feed

Deploy an ephemeral fake MCP server to detect adversarial probing

Get attack patterns observed by all active honeypots

Tear down a specific honeypot and retrieve captured data

List all active and destroyed honeypots with summary

Initiate an automated trust handshake with another AI agent behind Guardian

View all active trust relationships and session details

Immediately terminate a trust relationship

List all registered agents in the trust registry

Get overall status of all agentic AI features including metrics, scheduler, and task queue

Compute an A+-F trust score for an MCP server across 8 security dimensions (like SSL Labs for MCP)

Scan MCP tool responses for PII, credentials, sensitive paths, and data exfiltration

Run MCP server certification (Bronze/Silver/Gold/Platinum)

List MCP servers in the local certification registry with level and expiry

Verify a server certification attestation (JWS) and level

Declare session intent and allowed tools for intent-binding enforcement

Run MCP protocol fuzzer — test defenses against malformed JSON-RPC, overflow, injection

Check SLA compliance — p50/p95 latency, error rate, circuit breaker state per tool

Execute an incident response playbook (prompt_injection, credential_leak, shell_injection)

Get agent reputation score — Trusted/Standard/Suspicious/Blocked tier with bypass rate and entropy

Analyze MCP server config and get A-F hardening grade with one-click recommendations

Detect agent-to-agent collusion patterns (recon-then-exploit, coordinated exfil, token sharing)

Explain MCP Guardian policy YAML in plain English for compliance stakeholders

Convert a natural-language security goal into a draft YAML policy rule (requires approval before enforce)

Query decentralized MCP server reputation (8-dimension consensus score)

Compute cyber insurance ALE (Annualized Loss Expectancy) for an MCP server

Thompson Sampling — run Bayesian bandit trust sampling for an agent (Beta posterior, exploration/exploitation)

Contextual Bandit (LinUCB) — select optimal policy action (enforce/relax/skip) based on context

SARSA — adaptively tune rate limit, latency limit, or confidence threshold via reinforcement learning

REINFORCE — use policy gradient to select optimal fuzzer mutation strategy

概要

What is MCP Guardian?

MCP Guardian is a runtime security, cost governance, and health monitoring proxy for MCP infrastructure. It uses a three‑layer detection engine (regex → schema → LLM), a YAML policy engine with hot‑reload, and provides enterprise observability.

How to use MCP Guardian?

Deploy MCP Guardian as a proxy between MCP clients and servers. Define security and cost policies using YAML files that support hot‑reload. Monitor security events, token costs, and server health through the built‑in web dashboard and Prometheus metrics.

Key features of MCP Guardian

  • Three‑layer detection engine (regex, schema, LLM)
  • YAML policy engine with hot‑reload
  • 50+ secret patterns and Shannon entropy detection
  • AST command validation and JWT algorithm pinning
  • Default‑deny fail‑closed security model
  • Token cost tracking via tiktoken
  • Health monitoring with circuit breakers
  • Prometheus metrics and web dashboard

Use cases of M

コメント

「開発者ツール」の他のコンテンツ

FARPY

mangomunchr

Verified distributed compute for humans, applications, and agents. Submit Blender, Octane, and video workloads, route them across qualified GPUs, track asynchronous execution, download finished artifacts, and verify FARP

1

Foremerge

naw103

Catch intent conflicts before code conflicts. The open-source coordination protocol for coding agents, built above Git.

50

prodready

Innovate d.o.o. (ProdReady)

Free enterprise-grade due diligence for the app you built, run by your own coding agent: 118 read-only checks plus an 18-dimension review, scored 1-5 with file-and-line evidence and a report page behind an access code. N

Casefile

azimov777

One case file per task — shared by every session, agent and person that touches it. Self-hosted MCP server and live board. MIT.

3

GripForge

gripforgeai

The tools your AI needs to make your games. Turn prompts into production-ready game assets — animated characters with their weapons attached, seamless textures, terrain, VFX, HUDs — and playable game kits Unity, Godot, U

1
広告掲載