AgentGuard.Onnx
ONNX-based ML classifiers for AgentGuard - offline prompt injection detection. Includes the bundled StackOne Defender multi-head model (fine-tuned MiniLM-L6, minilm-multihead-v5, ~22 MB, calibrated dual-head with aux veto) which requires no separate download. Also supports the optional DeBERTa v3 model (download separately via eng/download-onnx-model.sh). See https://github.com/filipw/AgentGuard for details.
Install
dotnet add package AgentGuard.Onnx --version 0.6.0
Install-Package AgentGuard.Onnx -Version 0.6.0
<PackageReference Include="AgentGuard.Onnx" Version="0.6.0" />
Frameworks
Dependencies
Adoption guide
Assess this package against its supported frameworks, license, maintenance signals and dependency graph before making it a project standard.
Compatibility metadata lists 1 target framework; confirm the exact target used by your application.
How to read this guidance
Sources: published NuGet v3 registry metadata and explicit NuBrowse editorial guides. Refresh: package metadata is refreshed from NuGet when the page loads. Limit: this is not a security, legal or compatibility guarantee; validate the selected version in your project.
Useful next steps
Move from package metadata to a concrete selection or review workflow.