Log-To-Leak: Prompt injection attacks on tool-using LLM agents via MCP

# Log-To-Leak: Prompt injection attacks on tool-using LLM agents via MCP

https://gcdradar.vercel.app/incidents/log-to-leak-prompt-injection-attacks-on-tool-using-llm-agent · AIR-2026-1606 · closed 2026-11-11

- Lane: Research
- Severity: 1 (Research / proof of concept): Research demonstration of an attack vector against LLM agents.
- Fix status: Fix status unknown
- First reported: 2026-10-28
- AI role: Attack on AI
- Products: Model Context Protocol

## Summary

Researchers introduce Log-To-Leak, an additive post-hoc logging attack that exploits natural-language metadata in the Model Context Protocol. The attack induces unauthorized tool calls and targets user queries and agent responses.

## What to do

No action recorded; read the sources below and the vendor's advisory.

## Latest updates

- 2026-10-28: Log-To-Leak: Prompt Injection Attacks on Tool-Using LLM Agents via the Model Context Protocol (https://arxiv.org/abs/2610.36108)

## Sources

- 2026-10-28 arXiv: Log-To-Leak: Prompt Injection Attacks on Tool-Using LLM Agents via the Model Context Protocol https://arxiv.org/abs/2610.36108

Verify against the vendor's advisory before changing production. Data CC BY-SA 4.0, https://gcdradar.vercel.app.