Lucid Train

Lucid Train is built for developers who want a repository’s architecture to shape coding-agent work. Its strongest fit is a systems-design-first workflow; it is less suited to teams seeking a general-purpose diagramming canvas.

Overview

The product links codebase mapping to implementation: an agent produces a semantic graph, and the ELK layout engine calculates diagram positions and orthogonal edge routes. The resulting diagram can guide a coding agent as a specification, making architecture part of the work rather than a separate illustration.

Proposed edits appear as diffs, and commands are assigned risk tiers, with catastrophic commands refused. These safeguards make changes easier to inspect, but developers still need to judge whether a proposed change is sound.

Key features

• Diagrams tied to code: Diagrams are Excalidraw scenes that can be exported to SVG or PNG, or converted into Markdown specifications. Mermaid support and diagram import extend the workflow beyond diagrams generated from a repository.
• Choice of model: Lucid Train works with local Ollama models, OpenRouter, other OpenAI-compatible endpoints and hosted cloud models. Local processing of repository files and support for local models or users’ own keys give developers options for keeping work tied to a chosen provider.
• Connector and MCP options: The Connectors tab includes GitHub, Vercel, Railway and Grafana, and can install arbitrary stdio MCP servers. The stdio-only restriction rules out direct connections to remote HTTP and SSE servers, which may constrain teams whose tools expose those interfaces.
• Local credential handling: API keys are stored in the Lucid Train configuration directory or OS keychain, not sent to the maker. The maker says it collects no telemetry, analytics, prompts, code or usage data from local or bring-your-own-key sessions. Prompts and code go to the selected model provider when one is used; with a local model, they go nowhere.

Pricing

Lucid Train uses a freemium model, with a free CLI and a paid native desktop app.

• CLI — free, forever: Includes the full agent, local models or users’ own API keys, and open-source software, with no account needed. It is the practical choice for developers who want the agent workflow without the native app’s design, diagram, audit and dashboard tools.
• Lucid Train — $2.50 a month, billed $30 a year: Adds the native app and its Design, Diagram, Audit and Dashboards tools to everything in the CLI. A 7-day free trial requires no card up front. The annual billing term and desktop license limit matter: one license covers up to three machines, and desktop sign-in is required.

The free tier preserves the core agent workflow, so the paid plan makes most sense for users who need the desktop tools. Developers wanting a desktop license across more than three machines should look elsewhere.

Platforms

Lucid Train supports Linux, macOS and Windows, with a desktop app and CLI. Its desktop license covers up to three machines.

Who it's for

Lucid Train suits developers who want to map a repository and use that diagram to direct a coding agent, especially those who value model choice and local handling of repository files. The free CLI is a strong starting point for that workflow. It is a weaker fit for people who need direct remote HTTP or SSE MCP connections, or simply want a general diagramming tool without the coding-agent focus.

Pros and cons

Pros

• Connects repository architecture diagrams directly to coding-agent specifications.
• The free CLI includes the full agent and supports local models or users’ own API keys without requiring an account.
• Diff-based edits and risk tiers, including refusal of catastrophic commands, help make agent activity reviewable.
• Supports SVG, PNG and Markdown outputs, with local credential storage and several model options.

Cons

• MCP connections are limited to stdio; remote HTTP and SSE servers cannot connect directly.
• The desktop app requires sign-in for licensing and one license covers only up to three machines.
• The design, diagram, audit and dashboard tools require the annual paid plan; the free CLI does not include the native app.

Alternatives

• Creately is worth considering for collaborative canvas work: its free plan includes unlimited canvases and collaborators, with a 45-item limit per canvas.
• InfraSketch is another free option to compare for diagramming.
• Nimbalyst may suit readers who prefer an MIT-licensed desktop and iOS app with source code available on GitHub.
• OpenCharts offers a free plan with 500K AI credits, one of each project type, one Code Canvas project, 10 GB of Drive storage and all export formats.
• AI Diagram Maker is an alternative for web-based AI diagram creation, with a free plan that includes 50 AI credits per month and code, SVG or PNG export.
• AI Flowchart is a web-based alternative with a free plan that provides daily credits and includes ads.
• Archflow offers a free web plan for project diagrams and user stories, with one project and three diagrams per project.
• ArchGPT is another freemium web option, with a free plan that limits credits, history, saved diagrams and prompt length, and watermarks exports.
• AI Diagram Generators and AI Wireframe Generators are broader directories to explore for other tools.

Verdict

Choose Lucid Train if repository architecture should guide coding-agent work: its diagram-to-specification workflow, full free CLI and flexible model choices are the main reasons to use it. Look elsewhere if you need remote HTTP or SSE MCP servers, or a desktop license covering more than three machines.