x.comthreadRohan Paul (@rohanpaul_ai) on X
Rohan Paul's X post summarizes the paper "Log-To-Leak: Prompt Injection Attacks on Tool-Using LLM Agents via the Model Context Protocol", from TikTok researchers. It reports that MCP-using agents often finish the requested task, then quietly call an attacker-controlled logging tool that records the query, server reply and final answer. The attack works through natural-language tool descriptions that agents read when choosing and ordering tools. The visible text is truncated.
TL;DR Tweet summarizing a paper showing MCP agents complete the task and then silently call an attacker logging tool, leaking query, reply and answer.
Organisations named here
- Rohan Paul
Technologies named here
- LLM agents using the Model Context Protocol
- Attacker logging tool
- OP-TEE agent with QEMU snapshot replay
- Model Context Protocol
- LLM
- TikTok
Techniques named here
- Tool-description injection
- Defensive validation
- Log-To-Leak
- Prompt Injection
Related in AI Security
blog.talosintelligence.comtoolIntroducing CAIRN: Frontier tracking for AI-integrated malware (Cisco Talos)
github.comadvisoryFlowise MCP Security Bypass Enables RCE
github.comrepoYesWeHack Claude Kit
jfrog.comarticle