An architect's dashboard for a codebase being changed faster than you can read it.
A local-first, deterministic code-architecture visualizer.
scan -> generate -> publish
Scan a repository, build a canonical graph, publish a self-contained static site
into .topo/. No hosted service, no telemetry, no account.
The goal is to put a software engineer in the human-in-the-loop position: every claim the tool makes about a codebase is backed by evidence you can follow, and everything a human adds is recorded as such.
Status: pre-implementation. No code has been written yet. The design is being driven from evidence rather than opinion — see below.
The product requirements live in issue #1. That issue governs; anything here that contradicts it is wrong.
Design decisions are made against measurements taken on real repositories
(mermaid, wezterm, d3, and others), not from assumption. Several settled
decisions have been overturned by those measurements.
.skill-log/ holds the append-only chronicle of that work: what was measured,
what it showed, what was decided, and what was later retracted. It is committed
deliberately. When the reasoning behind a decision is unclear, the chronicle is
the record — not memory, and not this README.
MIT. See LICENSE.