diff --git a/SillyTavern.md b/SillyTavern.md
new file mode 100644
index 0000000..e14d3b0
--- /dev/null
+++ b/SillyTavern.md
@@ -0,0 +1,28 @@
+# Running on SillyTavern
+
+To run TabbyAPI on SillyTavern, a few more configuration changes are required on SillyTavern's side.
+
+Make sure to get your TabbyAPI **api key** from `api_config.yml`. The admin key is not useable on SillyTavern.
+
+Inside SillyTavern's `config.conf`, add the following under `const requestOverrides`:
+
+```json
+{
+    hosts: ["127.0.0.1:5000"],
+    headers: {
+        Authorization: "Bearer <Your API key>"
+    }
+}
+```
+
+This configuration assumes you're using the default config.yml. If you have customized it, make sure to replace `127.0.0.1:5000` with whatever you set the IP and port to in your config.
+
+What this does is intercepts all your oobabooga requests and appends this API key on top of them for security purposes.
+
+Now, inside SillyTavern, just connect to your TabbyAPI URL using oobabooga's text completion API option!
+
+The fetching of models isn't accurate since ooba uses the top-most model in tge model directory (I am not sure why this is the case) to decide which model is loaded. Please make sure to **load a model** beforehand!
+
+Completions work without an issue, just know that all samplers in oobabooga's slider panel won't work in TabbyAPI yet.
+
+Hopefully, Cohee and Ross (or myself) will add full support for TabbyAPI in the not too distant future.