Skip to content

REFERENCE

Glossary

These terms have the same meaning throughout the Panolayer documentation.

Last updated

Read as Markdown

agent graph

The complete structure of an agent’s steps and connections. It can contain repeated behavior, branches, and checks.

Read more.

agentic engineering

Developing software with agents carrying out tasks while you review their work. You remain responsible for deciding what reaches your branch.

Read more.

app

A project Panolayer detects in your repository, such as a web frontend or an API service. A repository with several apps, like a monorepo, contains more than one. Plans count active apps.

Architecture

Panolayer’s view of your parsed source and its relationships. Coverage varies by language and repository.

Read more.

architecture map

The functions, modules, dependencies, and callers visible in Architecture. It is derived from source rather than a model-generated drawing.

Read more.

baseline

The state of the codebase before a change. Verification uses it to distinguish new failures from pre-existing ones.

Read more.

Changes

The view for reviewing changed files and committing them. Risk scores help you choose what to read first.

Read more.

container agent

An agent running inside Panolayer's hardened Linux container runtime. Every third-party agent runs this way.

Read more.

living documents

Documentation generated from Architecture. You regenerate it to reflect code changes; it does not regenerate automatically.

Read more.

loop

Repeated behavior inside an agent graph. In Panolayer, the loop is not the whole graph.

Read more.

Marketplace

The place to discover and install agents and graphs, and to publish your own. Third-party agents keep their own vendor sign-in and billing.

Read more.

Panolayer Agent

The built-in coding agent, found in My Agents. Its model use consumes the credits included in Free and paid plans.

Read more.

pre-existing failure

A failure present before the current change. Its label distinguishes it from a newly introduced regression.

Read more.

risk score

A model's assessment of a changed file's diff, to guide review attention. It is separate from verification and does not prove correctness.

Read more.

Risk Spectrum

The Changes view that orders changed files by risk. Read each file’s reason before deciding how much attention it needs.

Read more.

rule

An expectation about code behavior. Rules can be derived from code or added and corrected by you.

Read more.

verification

Checking a change against expectations about the code. Results can identify failures and affected callers, but do not cover every possible bug.

Read more.

Verification Engine

The part of Panolayer that evaluates changes against rules and architecture. Inspect its results to decide what needs investigation.

Read more.

worktree

A separate working checkout of a repository. It lets tasks work independently before their changes are merged.

Read more.