Introduction

architxt documentation

architxt: unlock the doc

Query your architecture, don’t hunt for it.

architxt is an IT architecture focused knowledge workbench that turns source documents into a structured, explorable contextual graph. Research suggests software practitioners spend ~32% of their day seeking and using information — for architects that time is split across documents, diagrams, wikis, tickets and ADRs. architxt collapses the hunt into one queryable architecture layer.

Quick startInstallationWhy architxt? →

Import documents and consistently tag entities1 / 3

Extract structure

Upload documents and automatically pull out entities, relationships, contexts, tags, and mental models.

How documents work →

Explore visually

Compose the contextual graph on the Views canvas, inspect detail panels, and export focused snapshots.

Explore Views →

Reflect and curate

Ask free-text questions against the graph and move answers into curated Workspace pages with focused sections.

Try Workspace chat →

Sync to Hindsight

Push documents, entities, and mental models into a Hindsight memory bank and keep them aligned over time.

Hindsight sync →

Start building

1PrerequisitesRuntime, Docling, LLM access, and optional Hindsight setup.2InstallationGet architxt running locally from the repository.3ConfigurationEnv vars, database, and Hindsight server settings.4Quick start tutorialEnd-to-end walkthrough from upload to curated page.

Other ways in

• Views – explore and compose the contextual graph visually.
• Chat – ask the architxt agent about your bank and move answers into curated pages.
• Context Patches – manage the living contextual graph layer on top of a Hindsight bank.
• Quick start video – watch the walkthrough on YouTube.
• Temporal Mosaic blog post – read the original announcement.
• For agents / LLMs – point an AI assistant at the architxt documentation.