diff --git a/README.md b/README.md
index 0ab2a0c..2491c21 100644
--- a/README.md
+++ b/README.md
@@ -20,6 +20,8 @@ Building an emitter is extremely easy. If your target format has determinisitc r
 
 Critically, computation of which lines of the source file are in "preformat mode" and which are not happens in the parse step; this information is included in each node of the abstract syntax list, which means a backend emitter can be *purely context-free* in its output of each node. The parser only iterates through the file contents once, but because the ASL has no shared state between nodes an emitter could theoretically run in parallel with an arity of the line count and thus complete in O(1) time (with arity = n).
 
+See the "design" file for more details on the implementation, including the type format for nodes and the list production.
+
 ## Why the name?
 
 I was scrolling through the gemini wikipedia article for inspiration and this phrase popped up.