Skip to content

Releases: HuTa0kj/vetix

Vetix v1.2.0 – Release Notes

Choose a tag to compare

@HuTa0kj HuTa0kj released this 31 Jul 10:01

Features

  • Supports JSON data export
  • Docker can be used for rapid detection
  • Skip duplicate scans for the same hash
  • You can choose the output directory yourself.

Improvements

  • The saving of log files has been removed, as it is redundant in most scenarios.

Vetix v1.1.3 – Release Notes

Choose a tag to compare

@HuTa0kj HuTa0kj released this 31 Jul 03:41

Bug Fixs

  • This resolves the issue of duplicate outputs from both the plugin's detected items and the large model's validation.

Improvements

  • The suggestion output has been removed. In most cases, a malicious SKILL usually means unavailable, so there is no need to modify its underlying nature.
  • The routing between nodes has been optimized to make it more efficient.
  • Similar detection categories were merged.
  • Plugins supports setting malicious category settings.
  • The display order of the headers in the output tables has been optimized, placing the risk names at the beginning.

Vetix v1.1.2 – Release Notes

Choose a tag to compare

@HuTa0kj HuTa0kj released this 30 Jul 10:28

Features

  • Supports creating plugin templates using the create --plugin command.

Improvements

  • Split the plugins to make each plugin more independent.
  • The detection classification is presented in more detail.

Vetix v1.1.1 – Release Notes

Choose a tag to compare

@HuTa0kj HuTa0kj released this 30 Jul 07:33

Features

  • Supports passing the output language via the command-line --language parameter.

Bug Fixs

  • Fixed the issue of incorrect file count calculation.

Improvements

  • The terminal display has been optimized.

Vetix v1.1.0 – Release Notes

Choose a tag to compare

@HuTa0kj HuTa0kj released this 29 Jul 08:27

Features

  • The brand-new plugin system supports the writing of custom Python plugins and offers more flexible file inspection strategies.
  • Plugins and large models are cross-validated; each detected risk item is re-evaluated using LLM combined with the actual file content.
  • Autonomous exploration and analysis agent with virtual sandbox
  • Supports controlling detection logic and criteria via SKILL.

Improvements

  • Use BaseChatOpenAI to be compatible with more OpenAI connectivity protocols.
  • The configuration file has been simplified, and different preferred models can be controlled through lite and pro.

Vetix v1.0.0 – Release Notes

Choose a tag to compare

@HuTa0kj HuTa0kj released this 07 Jul 09:25

Features

  • LLM two-stage analysis, pre-summarization of checks and behavioral detection
  • Flexible configuration of multiple models
  • Agent can track behavioral paths