Shrike Security

stdionone★ 1development

AI agent security scanner â protect LLM-powered apps from prompt injection, SQL injection, data exfiltration, and adversarial attacks via MCP.

Install

How to run this server

[ { "cmd": "npx shrike-mcp`", "imports": [] } ]

server path: shrike-security

Tools

What this server exposes

scan_prompt

Scan user/system prompts before LLM processing for prompt injection, jailbreaks, and other threats

scan_response

Scan LLM outputs before returning to user for leaked API keys, system prompt in output, and other data leakage

scan_sql_query

Scan SQL queries before database execution for SQL injection attacks like tautology injection

scan_file_write

Scan file paths and content before write for path traversal and sensitive data exposure

scan_command

Scan CLI commands before shell execution for command injection, reverse shells, and data exfiltration

scan_web_search

Scan search queries before execution for PII exposure in web searches

scan_a2a_message

Scan agent-to-agent messages before processing for prompt injection in inter-agent communication

scan_agent_card

Scan A2A AgentCard metadata before trusting for embedded injection in agent discovery and capability spoofing

check_approval

Check human-in-the-loop approval status and submit decisions for flagged actions

report_bypass

Report user-reported missed detections to feed ThreatSense adaptive learning

get_threat_intel

Get current threat patterns and intelligence including latest prompt injection techniques

reset_session

Clear session correlation state after resolving flagged patterns

Configuration

Environment & auth

authnone

envSHRIKE_API_KEY

envSHRIKE_BACKEND_URL

Resources