MCP security scan toolkit: scan, inspect, compliance check
MCP security scan, inspect, compliance check: three dashboard tools for shipping safer MCP servers. What each catches and how they fit a real workflow.
Static analysis on 12,750 n8n templates from n8n.io and GitHub. 716 expose pre-auth vulnerabilities. Six end-to-end demos: SSRF, SQL injection, RCE.
Thoughts, stories and ideas.
MCP security scan, inspect, compliance check: three dashboard tools for shipping safer MCP servers. What each catches and how they fit a real workflow.
Per-IP isn't enough for an AI gateway. How to size your MCP rate limits for DoS, free-tier abuse, and per-tenant fairness, with worked examples.
A 14-recipe Lua lambda catalog for AI gateways: response shaping, request preprocessing, error normalization. One-click install on MCP or LLM proxies.
n8n's MCP Server Trigger ships with a bearer token and per-workflow opt-in. Six months later you have ten endpoints, one token, and no idea who calls what.
MCP tool responses join the LLM context as trusted tokens. A leaked API key or customer email becomes part of the prompt the model reads and logs.
MCP security scan, inspect, compliance check: three dashboard tools for shipping safer MCP servers. What each catches and how they fit a real workflow.
Per-IP isn't enough for an AI gateway. How to size your MCP rate limits for DoS, free-tier abuse, and per-tenant fairness, with worked examples.
n8n's MCP Server Trigger ships with a bearer token and per-workflow opt-in. Six months later you have ten endpoints, one token, and no idea who calls what.
Field notes from the team. New CVEs we've seen, policy patterns that worked, and rough notes from the lab.
Thoughts, stories and ideas.