Summary
ClawGuard is a runtime security framework proposed to protect tool-augmented Large Language Model (LLM) agents from indirect prompt injection attacks. Tool-augmented LLM agents can automate complex multi-step real-world tasks, but adversaries can embed malicious instructions in tool-returned content, which agents incorporate into their conversation history as trusted observations. The paper identifies three main attack channels: web and local content injection, MCP server injection, and skill file injection. ClawGuard addresses this by enforcing user-confirmed rule sets at every tool-call boundary. This transforms unreliable, alignment-dependent defenses into a deterministic and auditable mechanism that intercepts adversarial tool calls before any real-world effect occurs. The paper (arXiv 2604.11790) is authored by Wei Zhao, Zhe Li, Peixin Zhang, and Jun Sun, and falls within the cs.CR and cs.AI research areas. It was released on April 13, 2026.
[Paper] ClawGuard: A Runtime Security Framework for Tool-Augmented LLM Agents Against Indirect Prompt Injection
Paper Overview
- Research areas: cs.CR, cs.AI
- Authors: Wei Zhao, Zhe Li, Peixin Zhang, Jun Sun
- Release date: 2026-04-13
- arXiv: 2604.11790
Abstract
Tool-augmented Large Language Model (LLM) agents have demonstrated impressive capabilities in automating complex, multi-step real-world tasks, yet remain vulnerable to indirect prompt injection. Adversaries exploit this weakness by embedding malicious instructions within tool-returned content, which agents directly incorporate into their conversation history as trusted observations.
This vulnerability manifests through three main attack channels:
1. Web and local content injection
2. MCP server injection
3. Skill file injection
The paper proposes ClawGuard, a novel runtime security framework that enforces user-confirmed rule sets at every tool-call boundary. This transforms unreliable, alignment-dependent defenses into a deterministic, auditable mechanism that intercepts adversarial tool calls before any real-world effect can occur.
---
*Auto-collected on 2026-04-15*
This page is an English static mirror generated for search and AI citation.
It may be a full translation or structured summary of the Chinese original.
Canonical interactive discussion lives on the Chinese page:
https://zhichai.net/topic/177618481