Ai

OpenAgent v4.4.0 Launches Automated Security Research Team

OpenAgent v4.4.0 Launches Automated Security Research Team

Security researchers just got a powerful new ally. The latest OpenAgent release, v4.4.0, introduces a /security-research skill that spins up an entire adversarial security team in a single command. Think of it as hiring five specialized auditors—without the wait or the cost.

What Changed

The new skill creates a five-member team operating in parallel: three vulnerability hunters targeting surface-level flaws, authentication and data exposure, and runtime/supply-chain weaknesses. Two proof-of-concept engineers then validate every finding. Critically, each result is calibrated by actual exploitability—there’s no severity rating without a demonstrable attack path. The system integrates CWE classification, OWASP WSTG/ASVS methodology, and CVSS v4.0 scoring out of the box.

To enable it, you set team_mode.enabled: true in your oh-my-opencode.jsonc configuration file. That’s it.

Why It Matters

Manual security reviews are slow and expensive. Most automated scanners generate noise. This approach flips the script: AI agents collaborate like a human red team, but at machine speed. The parallel execution alone cuts audit time by a factor of five. And because findings are tied to actual exploit paths, developers get actionable results—not a laundry list of theoretical risks.

For teams using OpenAgent, this turns a command-line tool into a security force multiplier. It’s practical, rigorous, and built on standards that enterprises already trust. That’s rare in the AI tooling space.

Official Source: https://github.com/code-yeongyu/oh-my-openagent/releases/tag/v4.4.0