Hugging Face's logo

armorer-labs
/
armorer-guard-semantic-classifier

Text Classification
Joblib
ONNX
scikit-learn
security
prompt-injection
secret-detection
armorer-guard
Model card Files
xet
 Community
2

Armorer Guard Learning Loop: local live feedback without model drift

#2
by armorerlabs - opened
Discussion
armorerlabs
 Armorer Labs org

We just pushed the Armorer Guard Learning Loop.

It is a Rust-native hybrid live learning layer for local agent security enforcement:

  • feedback adapts local enforcement immediately
  • global model improvements still go through reviewed, versioned retraining
  • no scanner network calls
  • no silent cloud upload
  • no poisoning-by-default

New CLI modes:

armorer-guard feedback-record
armorer-guard feedback-stats
armorer-guard feedback-export --reviewed-only

inspect / inspect-json now also include scan_id, model_version, and learning_version, so feedback can be tied back to a stable local scan without requiring raw text storage.

The local overlay can suppress eligible semantic false positives after feedback, or add local block/review matches. It deliberately cannot suppress credential or dangerous policy reasons like detected:credential, policy:credential_disclosure, or policy:dangerous_tool_call.

Repo/docs: https://github.com/ArmorerLabs/Armorer-Guard
Demo Space: https://huggingface.co/spaces/armorer-labs/armorer-guard-demo

Would love feedback from people building agents or eval harnesses: where would you put this loop: prompt ingress, retrieval ingress, model output, tool-call args, or all of them?
armorerlabs
 Armorer Labs org

Quick release update: the Rust CLI is now live on crates.io as armorer-guard v0.2.2.

Install:

cargo install armorer-guard --locked

This release includes the local Learning Loop (feedback-record, feedback-export, feedback-stats) so teams can adapt enforcement from reviewed local feedback without uploading prompts or silently mutating model weights.

Crate: https://crates.io/crates/armorer-guard/0.2.2
Repo: https://github.com/ArmorerLabs/Armorer-Guard
armorerlabs
 Armorer Labs org

Update from Armorer Labs: the classifier artifact repo has been refreshed.

What changed:

  • semantic_classifier_native.tsv was updated from the latest Armorer Guard runtime artifact.
  • Added semantic_classifier_profile_native.tsv, the Rust-native profile fallback used by higher-recall benchmark/strict profiles.
  • The model card/license metadata now reflects MIT.
  • The demo Space was rebuilt against the updated artifact repo.

This is still intentionally a local, lightweight scanner lane rather than a hosted inference endpoint. The runtime combines these artifacts with deterministic credential detection, context-aware policy reasons, MCP proxy enforcement, and local feedback overlay behavior.

If you are testing agent/tool-use safety, I am interested in examples where a local scanner should warn/review rather than block outright.
armorerlabs
 Armorer Labs org

Follow-up release status from Armorer Labs:

The model/artifact repo is current and now includes the refreshed native classifier TSV plus the profile fallback TSV. The GitHub runtime release v0.3.1 is also live: https://github.com/ArmorerLabs/Armorer-Guard/releases/tag/v0.3.1

Registry status is intentionally called out so people do not get confused:

  • Cargo CLI: armorer-guard 0.3.1
  • PyPI wrapper: latest installable is still 0.3.0 until package-token permissions are refreshed
  • npm wrapper: latest installable is still 0.3.0 until npm login/token is refreshed

The runtime code and model metadata are MIT-aligned now.

Sign up or log in to comment