Skip to content

Panolayer Studio · Graph Builder

Build a loop. Ship it as an agent.

Design how an agent works, not just what you tell it. Start from a built-in loop or a blank canvas, add checks and failure paths, and keep it private or publish it to the Agent Marketplace.

macOS 15 on Apple silicon · Free to start · No credit card required

Graph Builder · explicit paths and checks

A Panolayer graph with connected agent steps and separate failure and completion paths.

Full capture · scroll to inspect the details

Your prompt said run the tests. Your agent didn’t.

A prompt is a suggestion, and an agent can skip it. In Graph Builder, running the tests is a step the agent has to pass. If they fail, the loop takes the failure path you set instead of reporting done.

What is a loop

The agent does the work. The loop decides when it’s done.

A plain agent works until it decides it’s finished. A loop sets the steps, the checks after each one, and what happens when a check fails.

Agent does the work
Run the checks
Next step or done
Fix and try again
Back to the checks ↑
Stop and ask you
When retries run out

How it works

From a built-in loop to your own agent

01

Start from a built-in loop, or a blank canvas

Built-in loops include Ask, Coding Task, Coding Task + UI Testing and Figma MCP UI Build. Open one, change it and save it as your own, or start from scratch.

Graph Builder · start with a template

Graph Builder · start with a template

Built-in starter graphs and saved agent graphs in the Graph Builder library.

Full capture · scroll to inspect the details

02

Add the steps you need

Each step is a node on the canvas: agent steps, checks like tests, builds, type checks and browser checks, and approved command-line tools and MCP servers. Add conditions, waits, retries, recovery, and points where the loop stops to ask you.

Graph Builder · connect and configure steps

Graph Builder · connect and configure steps

Graph Builder walkthrough connecting steps and inspecting their configuration.

Full capture · scroll to inspect the details

03

Run it, or share it

Before a loop runs, Panolayer checks that every connection is valid, the permissions are in place, and every path ends in done or failed. A loop with no failure path won’t run. Panolayer Agent runs it step by step.

Every step has a way forward—and a failure path

Every step has a way forward—and a failure path

A saved Graph Builder canvas connects Start to a custom task and a script, with explicit routes to Failed and Done.

Full capture · scroll to inspect the details

Questions

Something else? Contact us and we’ll answer.

Do I need to write code?

No. You build loops by connecting nodes. Connecting your own command-line tools or MCP servers may take some setup.

What can a node do?

Run an agent, run checks, call approved tools or MCP servers, branch on a condition, wait, retry, recover from a failure, read or write project memory, or stop and ask you.

Who runs my loop?

Panolayer Agent, following the pass and fail paths you set.

Can I keep a loop private?

Yes. It stays private unless you publish it.

What does it cost to run a loop?

Each run uses Panolayer Agent usage, and every plan includes some each month. Longer loops with more steps and retries use more. See pricing.

Does a published loop run on my API key or the installer’s?

The installer’s.

Build the agent your project needs.

Free to start · No credit card required

How Panolayer handles your code

Next: Agent Marketplace