AgentFirewall
Project AI agents from unsafe actions before they execute.
Start new thread

AgentFirewall - Project AI agents from unsafe actions before they execute.

by•
Model-agnostic security firewall for AI agents. Evaluate tool calls, MCP, and API actions before execution — allow, block, or require human approval.Model-agnostic security firewall for AI agents. Evaluate tool calls, MCP, and API actions before execution — allow, block, or require human approval.

Add a comment

Replies

Best
Maker
📌
I built AgentFirewall because AI agents are becoming capable of taking real actions, but developers still need control over what those agents are allowed to execute. AgentFirewall evaluates tool calls, MCP requests, and API actions before execution and can allow, block, or require human approval. The goal is to make autonomous AI agents safer without taking away their usefulness. I’d love feedback from developers building agents and automation systems.