๐Ÿ›ก๏ธ Agent Guard

Whether the model resists or complies, unauthorized actions are blocked.

Demo Explainer: Attacker Output Differences

Need help staying on top of your agents?

Let's talk.

๐Ÿ” Free Scanner

Security scanner for AI agent configurations. 20 rules, A-F scoring, CI/CD ready.

npx @musashimiyamoto/agent-guard scan .
View on npm โ†’

๐Ÿ”ฌ Free Cost Saving

Cut AI costs 50-80% with semantic retrieval + compression. FAISS + LLMLingua-2.

pip install tokenshrink
View TokenShrink โ†’