Radware Adds Agent Security for Claude Code

SiliconANGLE reported on July 7, 2026 that Radware expanded its Agentic AI Protection product with compliance reporting, deeper visibility into AI agent activity, and protection for developer-hosted agents including Anthropic's Claude Code. The report says the update maps agent activity, interactions, and dependencies, and adds audit-ready reporting aligned to ISO 42001, the EU AI Act, and the NIST AI Risk Management Framework. "Organizations are deploying AI agents across increasingly complex environments, creating new requirements for visibility, governance and security," said David Aviv, Radware's chief technology officer, in comments to SiliconANGLE. "These enhancements help organizations better understand agent behavior, support their compliance efforts and help extend protection to AI agents operating across both SaaS and local developer-hosted environments."

According to SiliconANGLE's reporting, the update lets security teams monitor agent behavior across conversations, govern which tools an agent can use, and guard sensitive data on developer machines. That matters because local coding agents often run with developer context: repository access, shell tools, build systems, tokens, and sometimes production-adjacent credentials. Agent governance that misses the endpoint can undercount the most sensitive tool calls. Radware's own product page confirms the Agentic AI Protection line combines visibility, governance, and real-time behavioral defense, though LDS could not independently verify the specific prompt-injection and jailbreak-control details from the vendor's own page directly.

The useful evaluation checklist is telemetry coverage, tool-control enforcement, data-loss safeguards, prompt-injection handling, and whether the product exports evidence into existing GRC, SIEM, endpoint, or secure-development workflows. A separate agent-only dashboard is less useful if security teams cannot correlate local agent behavior with identity, device posture, repository access, and incident response.

Buyers should look for independent testing, clear deployment boundaries, and integration details rather than only framework alignment claims. The stronger signal will be whether agent-security products can enforce controls without breaking developer workflows or creating another unmanaged endpoint agent.