diff --git a/MANIFEST.in b/MANIFEST.in
index 41b6710..5457f1c 100644
--- a/MANIFEST.in
+++ b/MANIFEST.in
@@ -1,3 +1,2 @@
 include examples/forecast.py tests/conftest.py tests/distribution.py
 include uv.lock .python-version
-exclude README.md
diff --git a/README.md b/README.md
index 9e591bd..7a96137 100644
--- a/README.md
+++ b/README.md
@@ -1,7 +1,7 @@
 <p>
   <picture>
-    <source media="(max-width: 600px)" srcset="./assets/readme/header-mobile.svg">
-    <img src="./assets/readme/header.svg" alt="ci-proto · Continuous Intelligence"
+    <source media="(max-width: 600px)" srcset="https://raw.githubusercontent.com/continuousintelligence/ci/main/assets/readme/header-mobile.svg">
+    <img src="https://raw.githubusercontent.com/continuousintelligence/ci/main/assets/readme/header.svg" alt="ci-proto · Continuous Intelligence"
       width="1200">
   </picture>
 </p>
@@ -52,7 +52,7 @@ ci.plot_forecast(history, prediction, model.quantiles, output="forecast.png")
 ```
 
 ![Five days of hourly bicycle rentals followed by ci-proto's next-day median
-forecast and nested quantile bands.](./assets/readme/bike_sharing_forecast.png)
+forecast and nested quantile bands.](https://raw.githubusercontent.com/continuousintelligence/ci/main/assets/readme/bike_sharing_forecast.png)
 
 **The next 24 hours, with uncertainty included.** The shaded region marks the forecast
 day. The curve shows the model's median forecast, surrounded by nested quantile bands.
@@ -73,7 +73,7 @@ covariates, practical uses of the quantiles, and forecasting through missing dat
 [![Mean scaled quantile loss on fev-bench versus parameter count for ci-proto and other
 forecasting models. Lower is better.][frontier]][frontier]
 
-[frontier]: ./assets/readme/ci_proto_pareto_frontier.png
+[frontier]: https://raw.githubusercontent.com/continuousintelligence/ci/main/assets/readme/ci_proto_pareto_frontier.png
 
 ## Talk to the team
 
@@ -367,7 +367,11 @@ requests focused, and run the checks above before submitting.
 ## Licence
 
 The code and model weights are released under the
-[ci-proto Code and Model Licence](LICENCE). Third-party components retain their upstream
-licences; see [NOTICE](NOTICE) for attribution and the Apache 2.0 licence text.
+[ci-proto Code and Model Licence][licence]. Third-party components retain their upstream
+licences; see [NOTICE][notice] for attribution and the Apache 2.0 licence text.
 The bundled example data is licensed under CC BY 4.0;
-see its [notice](src/continuousintelligence/_data/NOTICE).
+see its [notice][data-notice].
+
+[licence]: https://github.com/continuousintelligence/ci/blob/main/LICENCE
+[notice]: https://github.com/continuousintelligence/ci/blob/main/NOTICE
+[data-notice]: https://github.com/continuousintelligence/ci/blob/main/src/continuousintelligence/_data/NOTICE
diff --git a/pyproject.toml b/pyproject.toml
index aec7fe5..41b0968 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -4,9 +4,9 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "continuousintelligence"
-version = "0.1.0"
+version = "0.1.1"
 description = "Continuous Intelligence models"
-readme = { text = "Continuous Intelligence models.", content-type = "text/markdown" }
+readme = "README.md"
 license = "LicenseRef-ci-proto AND Apache-2.0 AND CC-BY-4.0"
 license-files = [
     "LICENCE",
diff --git a/uv.lock b/uv.lock
index 5bb031c..66a70da 100644
--- a/uv.lock
+++ b/uv.lock
@@ -320,7 +320,7 @@ wheels = [
 
 [[package]]
 name = "continuousintelligence"
-version = "0.1.0"
+version = "0.1.1"
 source = { editable = "." }
 dependencies = [
     { name = "apache-tvm-ffi", marker = "platform_machine == 'x86_64' and sys_platform == 'linux'" },