TRUSTDESC prevents tool poisoning in LLM applications by automatically generating accurate tool descriptions from code via a three-stage pipeline of reachability analysis, description synthesis, and dynamic verification.
Datasentinel: A game-theoretic detection of prompt injection attacks
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.CR 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
SONAR constructs a relational graph from entailment and contradiction scores to prune injected malicious sentences from LLM prompts while preserving context, achieving near-zero attack success rates.
citing papers explorer
-
TRUSTDESC: Preventing Tool Poisoning in LLM Applications via Trusted Description Generation
TRUSTDESC prevents tool poisoning in LLM applications by automatically generating accurate tool descriptions from code via a three-stage pipeline of reachability analysis, description synthesis, and dynamic verification.
-
A Sentence Relation-Based Approach to Sanitizing Malicious Instructions
SONAR constructs a relational graph from entailment and contradiction scores to prune injected malicious sentences from LLM prompts while preserving context, achieving near-zero attack success rates.