[ SLANG ][ AI CULTURE ][ MCP SECURITY ]
An attack where a tool exposed to an AI agent (via MCP or similar protocols) contains hidden malicious instructions in its description or behavior, manipulating the agent into taking unintended actions when it calls that tool.
REAL-WORLD EXAMPLE
"That MCP server had tool poisoning baked into the description field, invisible to the user but read in full by the agent."
LORE & ORIGIN
Documented in MCP and agent security research as a distinct attack class, exploiting the fact that agents often trust tool descriptions and outputs without independent verification.
FIRST SEEN
2026-06
PEAK POPULARITY
2026 (MCP / Agent Security)
CURRENT STATUS
Mainstream Gen Alpha Slang
Related To:Skill PoisoningTool Shadowing
EQUIVALENT CONCEPTS IN OTHER LANGUAGES
AURA IMPACT ⓘ
+5 AURA
TREND STATUS
↑ HIGH RISING FAST · HIGH
CULTURE CATEGORY
[ SLANG ][ AI CULTURE ][ MCP SECURITY ]
RELATED TERMS
MENTIONS OVER TIME
2023202420252026
RELATED SEARCHES
TOP PLATFORMS
TikTok
68%
YouTube
15%
Twitter / X
9%
Reddit
5%
Others
3%
WHEN DID YOU FIRST HEAR THIS?
CLICK AN OPTION BELOW TO CAST YOUR VOTE.
[ VOTE TO REVEAL COMMUNITY RESULTS ]