diff --git a/README.md b/README.md
index 6a6922e5..23caed1f 100644
--- a/README.md
+++ b/README.md
@@ -36,9 +36,3 @@ _Status: alpha_
 - [Source code](https://github.com/sourcegraph/opencodegraph)
 - [Docs](https://opencodegraph.org)
 - License: [Apache 2.0](LICENSE)
-
-## TODOs
-
-- item.ui and item.ai
-- add support for resolving annotations (because )
-- classnames -> clsx for consistency
diff --git a/web/content/docs/concepts.mdx b/web/content/docs/concepts.mdx
index 4d1e0083..b4ec32fb 100644
--- a/web/content/docs/concepts.mdx
+++ b/web/content/docs/concepts.mdx
@@ -67,3 +67,7 @@ Will OpenCodeGraph providers be super lightweight presentation-layer programs th
 ### Recording additional metadata
 
 How should annotations record other kinds of metadata to make it searchable and exportable? For example, the Storybook provider could add `{"storybook": true}` metadata in an annotation on all storybook files, which would make it possible to identify all storybook files in a code search application.
+
+### Multiple resolution passes
+
+Should annotations initially be returned with minimal information that can be quickly produced, and then resolved to provide additional detail (such as when the user interacts with the chip)?