Keep your architecture docs in sync with your code | Tecture

Docs that never go stale<br>Keep your architecture docs in sync with your code<br>Your coding agent writes the architecture model straight into your repo and updates it as the code changes — so the docs are reviewed in your PRs and never drift out of date.

01Install the skill<br>terminal — bashCopy<br>$ npx @tecture/skill@latest

02Run it in your coding agent<br>coding agentCopy<br>/ architecture-docs

03View it<br>VS Code extensionLocal browser<br>Get the VS Code extension<br>terminal — bashCopy<br>$ npx @tecture/core

→ opens the viewer at localhost:3000 — no editor required.

the skill writes the model · view it in VS Code or your browser<br>Add the skillView on GitHub

Watch the demo<br>How Tecture works<br>A short overview of Tecture mapping the Redis repo into a navigable architecture model.

Demo: Documenting Redis repo architecture<br>video · watch the demo

REC ●<br>demo · documenting redis repo architecture

The problem<br>Your architecture docs are out of date — and you know it<br>Architecture lives in your code, but the documentation lives somewhere else: a diagramming tool, a wiki, someone's head. The two drift apart the instant you ship.<br>FIG.01<br>Docs rot the moment you merge<br>Hand-drawn diagrams and wiki pages are stale by the next pull request. Nobody trusts them, so nobody updates them.

FIG.02<br>Onboarding takes weeks<br>New engineers reverse-engineer the architecture by reading code and asking around, because there's no accurate map of the system.

FIG.03<br>Refactors become guesswork<br>Without an up-to-date view of components and dependencies, every change risks breaking something nobody knew was connected.

How it works<br>Your agent writes the model. Tecture makes it navigable.<br>Tecture stores architecture as simple JSON and Markdown files inside your repo. Your coding agent writes the structure and descriptions, and you review them like any other code change. Open the same files in VS Code or the browser to explore your system as an interactive, multi-level diagram.

Your agent authors<br>Coding agent skill<br>Your agent reads the source and writes the architecture as JSON + Markdown, committed next to the code.<br>$ npx @tecture/skill@latest

writes<br>File-based model · in your repo<br>manifest.json diagrams/system-context.json descriptions/api-server.md

renders

You navigate<br>VS Code or browser<br>Opens those same files as an interactive, multi-level diagram you drill through — in the VS Code extension, or run npx @tecture/core to view it in any browser.<br>VS Code · or your browser

Quick tour<br>The architecture model, level by level<br>Real screenshots of Tecture mapping the Redis repo — the same model your agent generates for any codebase.<br>View 01 / 04<br>Start with the system context<br>See how your whole system fits together at a glance.

View 02 / 04<br>Drill into the containers<br>Every service, datastore, and dependency, and how they connect.

View 03 / 04<br>Descriptions that live with the code<br>Open any node for the docs your agent wrote.

View 04 / 04<br>Down to components & files<br>Drill inside a service to see its internal components.

01 / 04

What you get<br>FIG.01<br>Agent-authored docs<br>Your coding agent reads the source and writes the architecture model as JSON and Markdown.

FIG.02<br>Repo-native files<br>The model lives beside the code, so it can be reviewed, versioned, and updated through normal development workflows.

FIG.03<br>Interactive architecture viewer<br>Explore components, dependencies, and descriptions in VS Code or any browser.

Get started<br>Keep your architecture in sync — starting with your next commit<br>Add the skill to your coding agent and the architecture model is generated, reviewed in your PRs, and kept current as your codebase evolves.<br>MIT<br>Open source

npm<br>@tecture/core

PR-native<br>Reviewed like code

terminal — bash<br>$ npx @tecture/skill@latest

Add the skillView on GitHub