L

Lucid Train

Generate architecture diagrams from code repositories, turn them into specs for coding agents, and run fully offline with local models.

freemium
Last Updated 9/6/2026
Visit Website

About Lucid Train

Lucid Train is an AI tool that generates architecture diagrams from code: point it at a repository to map the system and turn the chart into a specification for coding agents. It offers a desktop app and a 6 MB Rust CLI that can run fully offline with local models, keeping code on your machine. Charts are saved as Excalidraw scenes and export to SVG and PNG or Markdown specifications. All edits appear as diffs, commands are risk-tiered, catastrophic commands are rejected, and dependency and MCP security audits cover nine ecosystems. It supports any OpenAI-compatible endpoint, with different models for different roles.

Best for

Developers who need to understand an unfamiliar codebase quickly
Engineers and teams using AI coding agents for daily development
Technical leads concerned about dependency and supply-chain security
Teams with strict code-privacy requirements that need to process private repositories offline

Best use cases

Generate an architecture diagram when taking over an unfamiliar repository to map components and dependencies quickly

Create an architecture specification before coding and hand it to a coding agent for implementation

Analyze private code offline or in an internal network so it never leaves the machine

Audit project dependencies and MCP configurations regularly to find potential security risks

Features

Code to architecture diagram

Point it at a code repository to draw an architecture automatically, then hand the chart to a coding agent as a specification.

Flexible export

Charts are Excalidraw scenes that export to SVG and PNG and generate Markdown specifications in one click.

Reviewable edits

Every edit is shown as a diff; commands are risk-tiered and catastrophic commands are rejected outright.

Dependency and MCP auditing

Built-in osv-scanner plus npm, cargo, and pip audits cover nine ecosystems and check MCP configurations.

Pros

  • Runs fully offline with no telemetry; code stays on your machine
  • Supports any OpenAI-compatible endpoint, with different models for different roles
  • Edits are shown as diffs, commands are risk-tiered, and catastrophic commands are rejected
  • Charts use Excalidraw scenes, export to SVG/PNG, and generate Markdown specs in one click

Pricing Plans

Desktop (monthly)

$3month
  • 7-day trial

Desktop (yearly)

$30year
  • Works out at $2.50/month

* Pricing information is for reference only. Please refer to the latest announcement on Lucid Train's official website.

FAQ

The command-line tool is free and open source; the desktop app costs $3/month or $30/year (about $2.50/month), with a 7-day trial. Model usage is not billed by the official service: use local models for free, or pay providers directly with your own API Key.
Yes. With a local Ollama model, it can run fully offline with no API Key or telemetry; code stays on your machine, and credentials are stored in the operating system keychain.

Tool Information

DeveloperLucid Train
Release Date9/6/2026
Pricingfreemium
Visit Website