Assessment results of the future functionality survey

[rsinding]

We recently conducted a vote on future feature development and received a lot of feedback from universities, GitHub, and the community. For the top-voted features, we have made the following evaluations:

1.Agent workflow security scanning (Dify/n8n/LangGraph, etc.) is planned for release in version V4, subject to scheduling.

2.Regarding the open-sourcing of jailbreaking algorithms, we plan to initially release a subset of jailbreaking attack operators in early December.

3.Regarding suggestions to optimize the speed and accuracy of MCP scanning, we have refactored and optimized the current user experience issues. The refactored code expected to be released in early December. In addition, we plan to introduce a new feature for remote dynamic scanning of the MCP server at the end of December, which will support a combined static and dynamic recognition module and automatic verification.

4.Malicious Prompt Word Detection, being more of a defensive feature and deviates from the red team tool's objectives, is temporarily suspended. We will continue to monitor and evaluate whether to in clude this feature in the future.

Other features with fewer votes will be continuously monitored for changes in user demand. If a clear demand trend emerges, we will restart the evaluation process.

[futurelighthouse]

Welcome feedback and more suggestions.