diff --git a/docs/data.md b/docs/data.md
index 468934e..c8642bf 100644
--- a/docs/data.md
+++ b/docs/data.md
@@ -60,3 +60,39 @@ curl -s https://tanbirramim.github.io/open-source-radar/data/issues.json \
 ```
 
 A snapshot of this data is also published as a dataset on Hugging Face: [TanbirRamim/open-source-radar](https://huggingface.co/datasets/TanbirRamim/open-source-radar), with an `issues` table and a `repositories` table.
+
+## Feeds
+
+Every language and topic has an RSS feed of beginner issues, so you can follow new ones in any feed reader.
+
+- Feed index: https://tanbirramim.github.io/open-source-radar/feeds/
+- One language: `https://tanbirramim.github.io/open-source-radar/feeds/<slug>.xml`, for example [python.xml](https://tanbirramim.github.io/open-source-radar/feeds/python.xml). The slugs are in `language_slugs` in `issues.json` (`cpp`, `csharp`, `go`, ...).
+- One topic: `https://tanbirramim.github.io/open-source-radar/feeds/topics/<slug>.xml`, for example [ai-ml.xml](https://tanbirramim.github.io/open-source-radar/feeds/topics/ai-ml.xml). The slugs are in `topic_feed_slugs`.
+
+## Python example
+
+Standard library only. It downloads `issues.json`, looks up each issue's repository language and prints the five newest beginner issues for one language.
+
+```python
+import json
+import urllib.request
+
+DATA_URL = "https://tanbirramim.github.io/open-source-radar/data/issues.json"
+LANGUAGE = "Python"
+
+with urllib.request.urlopen(DATA_URL, timeout=30) as response:
+    data = json.load(response)
+
+repos = data["repositories"]
+matches = [
+    issue
+    for issue in data["issues"]
+    if issue["level"] == "beginner" and repos.get(issue["repo"], {}).get("language") == LANGUAGE
+]
+matches.sort(key=lambda issue: issue["created"], reverse=True)
+
+print(f"{len(matches)} beginner {LANGUAGE} issues")
+for issue in matches[:5]:
+    print(f"{issue['repo']}#{issue['number']}: {issue['title']}")
+    print(f"  {issue['url']}")
+```