NewYorkers

Athanor — Codebase as a Living Organism

350+ tests covering parsing, graph operations, agent behavior, and ripple propagation. Full codebase understanding without context window limits. 3D visualization of code structure via a custom UI.

Timeline

Tech

The Problem

Approach

Parsed code into an AST graph using tree-sitter instead of flattening it into text.

Used deterministic BLAKE3 hashing to detect exact changes across the codebase.

Applied Louvain community detection to cluster code into semantic domains.

Designed stigmergy-based navigation: agents coordinate indirectly through pheromones, like ant colonies.

Implemented ripple-effect propagation so changes automatically cascade through dependencies.

Results

350+ tests covering parsing, graph operations, agent behavior, and ripple propagation. Full codebase understanding without context window limits. 3D visualization of code structure via a custom UI.

350+tests covering parsing, graph operations, agent behavior, and ripple propagation

Technology

Have a similar challenge?

Tell us about your project. We respond within 24 hours with a scoped plan.

Start a project