Equipping LLM agents with real-world tools significantly boosts productivity. However, transferring tool-use autonomy to agents also implicitly transfers associated privilege to the agent and the underlying LLM. The improper usage of these privileges can pose severe risks, including information leakage and infrastructural damage. While several benchmarks have been built to study agents’ security, they often rely on pre-coded tools and restricted interaction patterns. Such crafted environments differ substantially from the real-world, making it hard to assess agents’ security capabilities in critical privilege control and usage. Therefore, we propose GrantBox, a security evaluation sandbox specifically designed for analyzing agent privilege usage. By automatically integrating real-world tools and allowing LLMs to invoke genuine privileges, GrantBox assesses agents’ privilege usage under prompt injection attacks. Our results indicate that while LLMs exhibit basic security awareness and can block some direct attacks, they remain vulnerable to more sophisticated attacks, resulting in an average attack success rate of 84.80% in carefully crafted scenarios.