SecLLM/Playground/MCP security check

PLAYGROUND / 01

A closer look at
your MCP tools.

Inspect tool descriptions and server configuration before connecting an agent.

01 / INPUTJSON

What is your agent connecting to?

Try a sample
tools/list · tools[] · mcpServers0 / 500 KB

Input stays in this page until you clear it or leave. Reports can include snippets from your input; review them before sharing.

02 / REPORT

READY

Start with your configuration.

Paste JSON or choose a sample, then run the check. Findings include evidence and a suggested action.

↳ Tool description injection↳ Startup commands & secret exposure↳ Overlapping tool metadata
HOW TO READ THE REPORT

31 deterministic checks. Scores summarize matched signals, not exploit probability. No match does not establish safety. This check does not execute server code or verify authorization.

Read the MCP security starting points ↗