HiddenLayer Adds Runtime Security for AI Coding Agents

HiddenLayer announced Agent Harness Security in a press release, a runtime security product for AI coding agents. The product extends the company's AI Security Platform Runtime Security module and is available today as a standalone solution.

Agent Harness Security monitors coding agents as they read code, run commands, modify files, install dependencies, and interact with repositories. It integrates with each coding agent's native hook surface to detect prompt injection, sensitive data exposure, unsafe command execution, malicious dependency installs, and obfuscated payloads.

The product can redact secrets before they reach a model, steer agents away from poisoned tool responses, and stop unsafe actions inside the development workflow. It also gives security teams session context across prompts, tool calls, shell commands, file edits, and repository activity.

HiddenLayer said the product reports which controls are active for each coding agent and what enforcement level each platform supports. Actions can be marked as detected, redacted, blocked, or limited by the agent platform.