Skip to main content
Potpie parses every source file in a connected repository and runs inference on each node to build a knowledge graph, a structured representation of what every component does and how it relates to everything else. Ask, Build, and Debug all run against this graph.

What gets parsed

Every source file breaks down into typed nodes.
Node typeWhat it captures
Function / MethodSignature, parameters, return type, call sites, and the file it lives in
ClassFields, methods, inheritance chain, and every file that instantiates or extends it
Module / FileExports, imports, and the dependency relationships between files
Potpie indexes each branch independently, so queries always reflect the exact state of code on the selected branch.

How relationships are mapped

Relationships between nodes become typed edges in the graph.
Edge typeWhat it represents
CallsA function invoking another function
ImportsA file or module depending on another
ExtendsA class inheriting from another
ImplementsA class implementing an interface or contract
Every edge carries directionality. The graph records which component depends on which, giving agents a precise map of every dependency in the codebase.

How agents use it

Every answer is grounded in the graph. File paths and line numbers come directly from the indexed nodes.