The fastest way to try Archinsight is the online playground, where you can create and visualize an Insight model directly in the browser. For a local installation, use the Docker image. The CLI and VSCode extension bring validation, queries, navigation, and diagram generation into a local development workflow.
Try the online playground
The playground runs entirely in the browser and requires no installation. Open the Archinsight Playground, create an Insight tab, and begin describing your architecture.
Install the CLI
The Archinsight CLI package validates, queries, inspects, and renders Insight projects from the command line.
Install the latest release:
npm install --global @archinsight/cli@latest
Check the installed version:
archinsight --version
Validate all .ai files in the current project:
archinsight link . --format text
Use Archinsight with AI agents
Archinsight can generate a project-local skill that teaches AI coding agents how to understand, modify, and validate Insight architecture models.
For Codex:
archinsight skill init . --target codex
For Claude:
archinsight skill init . --target claude
A generic skill is also available for agents that support project-level instructions:
archinsight skill init . --target generic
The generated skill gives the agent the Insight language reference, built-in architecture framework, modeling workflow, query recipes, and examples. Agents can inspect the existing architecture, make changes, run graph queries, and validate their work through the same Archinsight linker used by the CLI and editor.
Restart the agent session after installing a native Codex or Claude skill so it can be discovered.
Install the VSCode extension
The Archinsight extension for Visual Studio Code adds Insight syntax highlighting, diagnostics, completion, project navigation, queries, and diagram previews.
Install it from the Visual Studio Marketplace or run:
code --install-extension archinsight.archinsight-vscode
Run Archinsight locally with Docker
The Archinsight Docker repository contains the web application for local and hosted installations.
Run a local, single-container instance:
docker run --rm \
--name archinsight \
--publish 127.0.0.1:3000:3000 \
--env ARCHINSIGHT_AUTH_TOKEN_SECRET="$(openssl rand -hex 32)" \
lonelylockley/archinsight:editor-ui-3.0.0
Open http://localhost:3000/app/playground after the container starts.
This configuration is suitable for local work and evaluation. A persistent multi-user installation can connect the same image to PostgreSQL and an authentication provider. Diagrams are rendered in the browser by default, so the external renderer service remains optional.
Build from source
The complete project is available in the Archinsight GitHub repository.
Clone the repository and build every release artifact:
git clone https://github.com/lonely-lockley/archinsight.git
cd archinsight
./gradlew clean dist
The build runs the project checks and creates the CLI package, VSCode extension, web application, and local Docker images.