Make your codebase make sense.
An architecture storybook for people building software with agents. Map one meaningful question at a time. Keep the explanation close to the code.
Start in your repository.
Node.js 22+ and Git. No account or hosted service.
npm install --save-dev @jdylanmc/topo-code@0.2.0npx --no topo init . --skills
npx --no topo serve .A working Home, before your first map.
From idea to architecture
Understand the code at a glance, faster than reading files. Your agent turns source evidence into a diagram; you review and refine the explanation.
Read the implementation.
Deterministic scanning supplies structural evidence. Your coding agent reads the source and helps explain what matters. Scanning never calls a model.
Author the explanation.
Trace a boundary, a request, a decision, or a lifecycle. Curate stories with your agent instead of accepting a wall of generated boxes.
Keep the evidence.
Follow claims back to source. Validate anchors, repair stale evidence, and share the storybook as static files.
A map is an explanation, not an inventory.
Fast-changing code needs more than another dependency graph. Topocode gives you a place to build and maintain a shared understanding: what a system does, why its boundaries matter, and what happens when things go wrong.
You and your agent own the narrative. Validation checks its structure and evidence; it does not certify that the explanation is complete.
Useful today. Honest about limits.
TypeScript and JavaScript analysis, source-grounded stories, a browsable Home, and static publishing. Version 0.2.0 also supports explicitly selected, partial Rust and Tauri analysis.
No compiler-complete Rust semantics, runtime reachability guarantees, or automatic architectural truth.
Read the support boundaries