Under development
Know exactly what your agent just did.
Bantay runs inline between your coding agents and your infrastructure. It logs every file, command, and outbound connection across every developer and CI runner — and stops the ones your policy hasn't allowed.
curl -fsSL bantay.ai/install | sh
soon
It asks before it's too late.
When the agent reaches for something outside your policy, Bantay holds the action and waits for you. No rollback, no postmortem.
Rules you can read.
One file in your repo. Version it, review it, share it with your team.
No action without an owner.
Every file read, command run, and connection opened is attributed to a person, a repo, a branch, and a named agent. Not "an agent did something on a laptop" — a specific action, by a specific agent, running on behalf of a specific engineer, at a specific commit. Reconstructable months later.
The same record, months later.
Every session your agents have run, across projects and machines, with the rules that actually fired. Read-only by design — approval stays in the terminal.
Plans are taking shape.
We're refining our plans and pricing. Full details will be available before launch.
Quickstart
brew install bantay-ai/tap/bantay
soon
bantay init
generate a policy
bantay watch
attach to a running agent session
Access is coming soon.
We're still building Bantay. Access will open when it's ready.