Make your network readable by agents

Agentic network operations built on eBPF:
fleet-wide, real-time, and incredibly cost-efficient

curl -fsSL https://yeet.cx | sh

Linux only  |  Manual install  |  Read the docs

New capabilities for your coding agents

With yeet, your live network traffic becomes the agent’s context for operating.

“yeet has been one of the biggest keys to our development success at Kairos. When you’re building trading infrastructure, you need tools you can trust to keep up, and yeet has done that from day one.

The technology powering it is seriously impressive, and it’s versatile enough that we keep finding new ways to use it across our stack. It’s become a core part of how we build, and I’d recommend it to any team working on serious infrastructure.”

Zayd Alzein, Co-founder & CTO
Kairos.trade, the prediction markets trading terminal

“yeet is an essential platform for us at Pickle in optimizing and monitoring network performance, an essential bottleneck that exists onboard every robot, between every robot, and across our growing footprint of distributed cloud and edge compute.

We prefer to build in-house, so we brought in yeet to let our engineers write their own systems for network performance issues that our existing telemetry could not explain. We find root causes in minutes.

Setup was fast, and yeet lets us replace millions of dollars of engineering and production expenses with tools our engineers build. Today, yeet is part of our daily workflow, continuously profiling the performance of our fleet.”

AJ Meyer, Founder & CEO
Pickle Robot, physical AI for supply chain automation

Use yeet for…

  • Fleet operations

    Improve cluster health via ambient limit checks across clusters, setting up ephemeral triggers where things look suspicious.

  • Incident triage

    Network incidents are the most challenging. Instead of digging through a thousand dashboards, generate specific views on the fly.

  • Sovereign troubleshooting

    Pre-install yeet in on-prem deployments so customers can self-diagnose network issues and relay answers back.

Use yeet when you don’t know where the problem is

Network problems live below the layer your tools were built to watch. yeet lets the agent go read that layer directly.

You don’t pay per event

Your kernel doesn’t bill you per event, and neither do we.

Free until you have a fleet. Three hosts, no charge.

curl -fsSL https://yeet.cx | sh
Monthly cost as events scale
$0
Splunk
$0
Datadog
$0
New Relic
$0
yeet
0events

It’s just Linux

yeet is a daemon on a Linux host. It runs the same on AWS, GCP and Azure, on Kubernetes nodes, in Docker or on bare metal.

Nothing about your networking changes: no sidecar, no proxy in the path, no CNI to swap. Your security or ours? Yours. yeet runs inside your account and your VPC, inherits the controls already there, and the data it reads stays on the host. Any agent that can run a command works with it.

Read the docs

  • AWS logoAWS
  • GCP logoGCP
  • Azure logoAzure
  • Kubernetes logoKubernetes
  • Docker logoDocker
  • Prometheus logoPrometheus
  • Grafana logoGrafana
  • Slack logoSlack
  • GPT logoGPT
  • Claude logoClaude
  • Jev logoJev
  • Gemini logoGemini

Security

The computation happens below the agent.

Your agent declares what it wants and gets the result. The intermediate computation (the raw kernel stream, the bytes a tool reads) stays inside the isolate and never crosses into the agent. Each tool is sandboxed the same way your Google Chrome sandboxes the web: a V8 isolate with no path to the network.

Build custom infrastructure

With access to runtime system data and yeet’s build primitives, your agents become capable of building a new category of infrastructure tools. From protocol debuggers to test infrastructure, every new tool in your stack is now custom-built to your needs.

Agentic network operations in under 60 seconds

curl -fsSL https://yeet.cx | sh

Linux only  |  Manual install  |  Read the docs