Patronus releases three open-weight classifiers for MCP tool-call security
PatronusProtect · reddit · 2026-07-21
Patronus released three open-weight classifiers for MCP tool-call security.
- Husky Sight classifies the tool type across 14 categories such as file, database, shell, API, secrets, and infra.
- Husky Paw classifies the operation: read, write, list, exec, or network.
- Husky Nose classifies data-flow risk, including labels like source:sensitive, source:untrusted, and sink:external.
- The models are designed to run before a tool call executes, enabling deterministic gates such as sensitive → external.
- Each model has a quantized edge build in ONNX INT8/INT4, around 96 MB, and is meant to run locally inside an MCP client without extra infrastructure.
More from coding & agent
- AWS on agent prompt caching: explicit + implicit patterns to slash token costs — AWS ML Blog · 2026-10-27
- He gave his AI agent a phone number and email — then hired it as social media manager — heyshrutimishra · 2026-10-27
- Firecrawl launches Alexandria company-data API and announces $75M Series B — devdigest · 2026-10-27
- 'Let an agent run wild on vague goals and it makes your code worse 9 out of 10 times' — eptwts · 2026-10-27
- LangChain to demo the in-house design agent its team uses daily, this Wednesday — LangChain · 2026-10-27
- Biotech Agent Series Day 8: A Notebook Review Agent to Auto-Audit Lab Records — nlarusstone · 2026-10-27