Tool poisoning
Definition
Tool poisoning is an attack that hides malicious instructions inside an MCP tool's description. The AI model reads tool descriptions as trusted context - a crafted description can instruct it to leak data or abuse other tools without the user seeing any of it.
Tool poisoning is prompt injection at the MCP layer. Countermeasures are reviewing tool descriptions on intake, versioning with mandatory approval on change - and a last line of defense before execution: critical tool calls wait for a human, no matter how they came about.
Related terms
tokyn.trail puts tracking, rules and human approvals in front of your MCP server - in 10 minutes, no code changes.
Learn more