Architecture
Navigate code you didn’t write.
Open a repository and Panolayer maps its architecture, with its modules, functions, classes, callers, and dependencies. The map is parsed from the source, not generated by a model, so the same code always produces the same map. You, your agents, and the Panolayer Verification Engine all work from it.
You’re three levels into a service, trying to work out what calls it.
A file tree shows you where code lives, but it doesn’t show you what depends on it. That gets even harder when an agent wrote the code and all you have is a thousand-line diff. Panolayer maps the dependencies, so you move through the system visually by what calls what, not by folder.
What you can do with it
01 / Architecture
Start with the whole system
Open a repository and the first thing you see is how it fits together including the services, what each one contains, and how they connect. It’s the picture you’d normally only get from someone who’s worked on the code for a while.
02 / Architecture
Drill down to what you need
Click into a service to see its modules, then into a module to see its functions. Diagram, column, and path views show the same structure three ways, so you can switch to whichever reads best for the code you’re in.
03 / Architecture
Read any function in context
Open getPrice() and you see what it does, what it takes and returns, what calls it, what it depends on, the rules that apply to it, and whether it currently passes verification. That’s everything you’d otherwise piece together from five files and a search.
04 / Architecture
See what a change will reach
Before you ask an agent to switch getPrice() from cents to dollars, check what calls it: the product page, the cart, and checkout. Panolayer follows the dependency graph from any function, so you know what a change can touch before anyone makes it.
05 / Architecture
Your agents see the same map
The agents you run can query this map through an MCP tool, so they can look up what calls a function before they change it, instead of working only from the files they were given. Verification checks every change against the same map, which means you, your agents, and verification all start from one picture of the system.
06 / Architecture
Documentation that keeps up
Living documents are generated from the architecture and stay in sync as the code changes, so the docs describe the system as it is now, not as it was when someone last updated them.
Panolayer remembers your project
Every new agent session starts from zero. You explain the same conventions, the same constraints, and the same reason nobody touches that one migration, again. Panolayer keeps a structured memory of your project across sessions including the decisions you’ve made, your conventions and constraints, your migration history, and how past bugs were fixed. The longer you work on a codebase in Panolayer, the less you have to repeat yourself.
Questions
Is the diagram generated by an LLM?
No. It’s parsed from the source. Models can add plain-language descriptions when you ask, but they don’t decide the structure.
Can I use the architecture map without running agents?
Yes. Open a repository and use it to find your way around code you didn’t write.
What languages are supported?
TypeScript, JavaScript, Python, Go, Rust, and Java, among others. Dynamic calls and generated code are harder to trace and may need a closer look.
See supported languagesDoes it work on a monorepo?
Yes.