import { Eye, Activity, ShieldQuestion, Zap } from 'lucide-react' export function Observability() { return (
Clarity

Watch it reach. Then decide.

You can't predict what your agent will access. Greywall makes every filesystem operation and network connection visible in real time — so you make informed decisions instead of guessing policies upfront.

{/* Monitor mode */}

Real-time violation monitor

Live violation stream
$ greywall -m -- claude
[14:23:01] DENY{' '} read{' '} ~/.ssh/id_ed25519
[14:23:01] DENY{' '} read{' '} ~/.aws/credentials
[14:23:02] DENY{' '} write{' '} ~/.bashrc
[14:23:03] ALLOW{' '} read{' '} ./src/index.ts
[14:23:03] ALLOW{' '} write{' '} ./src/utils.ts
{/* GreyProxy screenshot */}

GreyProxy dashboard

GreyProxy dashboard showing pending network requests with Allow and Deny controls
{/* Value props */}

Answer hard questions

What did the agent see? What did it access? What was denied? Greywall makes these questions answerable for any session, at any time, after the fact.

Move faster, safely

Once you can see what agents do and control it without breaking flow, you stop hesitating. Longer autonomous runs, more sub-agents, faster iteration on sensitive workloads.

Operate, don't hope

Observability is the difference between operating AI and hoping AI operates. Security becomes an enabler of velocity, not a constraint on it.

) }