LEO is an outliner, but by implementing a visually compelling directed acyclic graph (DAG) instead of a strict tree, it allows one to present multiple aspects of structure. Some "branches" of the graph can generate source code, others configuration files, others comprise documentation which includes analogs of "hard links" to portions of the code or configuration; various aspects of the subject can be pulled out and collected into coherent subsets independent of each other's structure.

Example: sendmail.mc generated by one tree, other trees within the graph contain source for the various .m4 files, while a documentation tree contains text but also nodes from the various files making the specific implementation immediately available for modification along with the documentation.

Example: trees within the graph show the structure of the source code to be generated (many languages have colorizer support, and any language with a reasonable comment syntax can be used); other trees may draw out nodes related to particular aspects, such as "all functions that invoke function foo()". Leo's own python source code is available in such a graph.

Example: a calendar with todo lists under each day, but individual items linked into global "Done" "Active" and Priority lists, and yet another section where the hierarchy reflects task dependencies.

Example: hosts characterized by types; types characterized by inheritance from other types and by makefile targets; different makefile components used by different OSs, and all those configuration files generated by trees within a graph, plus a tree for each host showing each type applied according to the type inheritance and the makefile targets along with the makefile dependencies.