radical.tools

What It Is

Radical.Tools is an MIT-licensed open-source architecture modeling tool. It is positioned not as a “diagramming tool,” but as a way to treat the architecture model as the documentation itself. It aims to solve the common problem of traditional architecture diagrams becoming outdated: diagrams are merely views derived from a unified model, so different levels and perspectives can remain consistent with one another.

Core Capabilities

The product uses a four-layer structure: the Metamodel defines the rules for allowed nodes, relationships, and properties; the Model stores the actual architecture graph that conforms to those rules; Views generate audience-specific representations from the same model; and Milestones provide named snapshots of model states. Built-in presets include C4, C4+DDD, and governance-enhanced versions that include ADRs and Fitness Functions. Users can also extend or replace the metamodel.

It offers a fairly broad range of view types, including Structure, Flow, Hierarchy, Table, Matrix, and Wiki. In other words, the same set of architectural facts can be represented as a system context diagram, container diagram, interaction steps, dependency matrix, editable governance table, or documentation page. For layout, it provides ELK, WebCola, and a simulated annealing pipeline aimed at reducing edge crossings. History features include snapshots, version comparison, and one-click restore.

AI, Pricing, and Openness

The main documentation clearly labels the project as MIT / open source, so it can generally be considered free and open source. However, no pricing information was found for commercial services or a hosted version. AI analysis supports Claude, OpenAI, Gemini, and local Ollama. Users can ask questions in natural language, identify gaps, or generate structured patches. Calls use the user’s own API key, and the documentation states they are not proxied through the platform. Its own API/SDK, plugin system, and import/export capabilities are not disclosed.

Pros, Cons, and Best Fit

Its main strengths are a clear model-driven approach, making it well suited to C4, DDD, architecture governance, and dependency analysis scenarios. The multi-view and milestone mechanisms are valuable for maintaining architecture assets over the long term. The limitations are that public information does not clarify multi-user collaboration, permissions, deployment dependencies, data storage, community activity, or enterprise support, so real-world validation is needed before adoption.

Access from China

The documentation does not state whether the website or related AI services are accessible from China, so this remains unknown. If it depends on OpenAI, Claude, or Gemini, domestic network access and payment may be restricted. Local Ollama can be considered to reduce external dependencies. Alternatives include Structurizr, PlantUML, Mermaid, Diagrams.net, Archi, and IcePanel.