RepoDAG — dependency graph for any repository

Draws a repository's code imports, CI pipeline and third-party packages as left-to-right dependency graphs, in your browser.

RepoDAG draws the structure of a source code repository as a graph you read from left to right. Open a repository on GitHub, click the toolbar icon and choose "Map this repository". A new tab downloads the source archive, reads it in your browser and shows three graphs. Code imports shows which files and folders depend on which. Everything sits to the left of the things it depends on, so entry points end up on the left and shared building blocks on the right. CI pipeline shows the jobs of the repository's workflows in the order they run, including jobs that wait for other jobs and workflows that start other workflows. Third-party packages shows each dependency file found in the repository and the packages it asks for directly, with the versions requested. A package used in several places appears once, so shared dependencies stand out. Packages needed only for development and tests are marked and can be hidden. Loops are found and listed. Choose one to see exactly which files depend on each other in a circle. The edge that closes a loop is drawn in red and dashed, pointing against the flow. Large repositories start grouped by folder so the picture stays readable. You can switch between files, packages and folders at a depth you choose, search by name, and click anything to see what it depends on and what uses it. The current view can be saved as an image, as text for common diagram tools, or as data. It also works without GitHub: drop a zip file or pick a folder from your computer, and it is read in the tab without being uploaded. Privacy: the only network request is the download of the repository archive from GitHub, made when you ask for it. There are no analytics, no accounts and no server. Limits: imports are found by reading source text, not by running a compiler, so unusual build setups and imports created at run time can be missed. The page tells you how many imports it found for each language and how many it could not match to a file, so you can judge the result.