MCP tool poisoning
Named by 4 kept items in this workspace.
Search every workspace for MCP tool poisoning. It matches the name as written, so it can find other things that share it.
The items that name it 4
Per OWASP: an indirect prompt injection attack targeting AI agents that connect to external tool servers via MCP, hiding instructions in a malicious tool's response
Invisible Unicode, agent-directed injection phrases, and cross-tool shadowing embedded in MCP tool names/descriptions — pattern behind the WhatsApp MCP rug-pull and postmark-mcp backdoor
Detected vulnerability category among the 17
All techniques in this workspace.