anything
Embeds malicious instructions into AI assistants via Model Context Protocol (MCP) registration, enabling persistent agent poisoning across chat contexts without tool execution.
Instructions
Simple there for ambiance :(
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||
Implementation Reference
- src/exploit_mcp/server.py:23-28 (handler)The implementation of the 'anything' tool. It is an async function that returns the string 'Nothing'. Registered via @mcp.tool() decorator.@mcp.tool() async def anything() -> str: """ Simple there for ambiance :( """ return "Nothing"