In development

Plug-and-play HTTPS egress filtering for AI agents

Bullfrog controls and monitors what your AI agents can reach on the network, running on any Linux infrastructure. It sees inside encrypted traffic to stop risky actions and data leaks, with no proxy to deploy and no certificates to install anywhere.

Occasional updates on progress and early access. Nothing else.

No proxy, no certificates

Every other approach to inspecting encrypted traffic means standing up a proxy and installing a custom certificate on every machine that talks through it. Bullfrog needs neither. It installs and starts working.

Blocking a domain isn't enough

Your agents need the same shared services everyone else uses, so a domain allowlist has to let the whole thing through. Bullfrog looks at what's actually being requested and decides from there.

Anywhere Linux runs

Agent sandboxes, build runners, servers, containers. You set the rules in one place and they apply everywhere.