Agent Security MCP Server

Security scanning, prompt injection detection, secret leak detection, and agent permission auditing for AI agent workflows. Built on the Model Context Protocol (MCP).

Tools

Tool	Description
`scan_mcp_config`	Scan MCP server configurations for security issues (dangerous commands, exposed secrets, network exposure, container misconfigs)
`detect_prompt_injection`	Analyze text for prompt injection attempts across 7 attack categories with context-aware risk scoring
`validate_scope_contract`	Check if agent actions comply with scope contracts (tool allowlists, file access, boundary constraints)
`scan_secrets`	Detect leaked API keys, tokens, private keys, database URIs, and credentials in text or code
`audit_agent_permissions`	Audit agent configurations against role-based expectations and flag principle of least privilege violations
`generate_security_report`	Generate comprehensive security assessment reports with prioritized remediation plans
`detect_tool_poisoning`	Analyze MCP tool definitions for hidden instructions that could manipulate agent behavior (OWASP Agentic Top 10)

Resources

Resource	URI	Description
OWASP LLM Top 10	`security://owasp-llm-top10`	OWASP Top 10 for LLM Applications (2025)
MCP Security Checklist	`security://mcp-security-checklist`	Security checklist for MCP server deployments

Installation

cd agent-security-mcp
npm install

Usage

As a standalone server

npm start

In Claude Desktop / MCP client configuration

{
  "mcpServers": {
    "agent-security": {
      "command": "node",
      "args": ["/path/to/agent-security-mcp/src/index.js"]
    }
  }
}

With npx (after publishing)

{
  "mcpServers": {
    "agent-security": {
      "command": "npx",
      "args": ["@asl-throne/agent-security-mcp"]
    }
  }
}

Detection Coverage

Prompt Injection (7 categories, 20+ patterns)

Instruction Override -- "ignore previous instructions", "disregard all rules", "new instructions:"
Identity Manipulation -- "you are now", "pretend you are", "act as", DAN/jailbreak
System Prompt Extraction -- "repeat your system prompt", "show your instructions"
Data Exfiltration -- "send this to", "post to webhook", "email everything to"
Delimiter Attacks -- ```system, [INST], <|im_start|>system, XML tag injection
Encoded Injection -- Base64 payloads, unicode zero-width characters, hex escapes
Privilege Escalation -- "sudo mode", "disable safety", "bypass filters"

Secret Detection (25+ patterns)

AI Provider Keys -- OpenAI (sk-), Anthropic (sk-ant-)
Cloud Credentials -- AWS (AKIA*), GCP (AIza*), Azure connection strings
Source Control -- GitHub PATs (ghp_, github_pat_), OAuth tokens (gho_*)
Payment -- Stripe live/test keys (sk_live_, sk_test_)
Communication -- Slack tokens/webhooks, Telegram bot tokens
Database -- PostgreSQL, MongoDB, MySQL, Redis connection URIs
Cryptographic -- RSA/EC/OpenSSH private keys, generic PEM blocks
JWT -- JSON Web Tokens
Generic -- api_key=, secret=, password=, .env file patterns

Permission Audit (6 role profiles)

Researcher -- Read + search + web only
Analyst -- Read + search only
Developer -- Read + write + execute
Reviewer -- Read only, no network
Orchestrator -- Read + write + task spawning
Monitor -- Read only, no network, no write

Pricing

Plan	Price	Servers	Features
Free	$0	1 server	Single scan, basic report
Starter	$49/month	3 servers	Continuous scanning, weekly reports
Pro	$199/month	20 servers	Real-time alerts, CI/CD integration, Slack notifications
Enterprise	$799/month	Unlimited	Custom policies, EU AI Act compliance reporting, SSO, dedicated support

Requirements

Node.js >= 18.0.0
@modelcontextprotocol/sdk >= 1.12.0

License

MIT

Agent Security MCP

Quick Install