diff --git a/GPT/beta-commands/beta-gpt.talon b/GPT/beta-commands/beta-gpt.talon
index f12e0007..0c033530 100644
--- a/GPT/beta-commands/beta-gpt.talon
+++ b/GPT/beta-commands/beta-gpt.talon
@@ -11,3 +11,7 @@ model blend clip:
     destination_text = edit.selected_text()
     result = user.gpt_blend(clipboard_text, destination_text)
     user.gpt_insert_response(result, "")
+
+# Pass the raw text of a prompt to a destination without actually calling GPT with it
+model pass <user.modelPrompt> [{user.modelDestination}]:
+    user.gpt_insert_response(modelPrompt, modelDestination or "")
diff --git a/GPT/gpt.py b/GPT/gpt.py
index 955a4009..0084af50 100644
--- a/GPT/gpt.py
+++ b/GPT/gpt.py
@@ -154,6 +154,11 @@ def gpt_apply_prompt(
             else text_to_process
         )
 
+        # Apply modifiers to prompt before handling special cases
+        match modifier:
+            case "snip":
+                prompt += "\n\nPlease return the response as a textmate snippet for insertion into an editor with placeholders that the user should edit. Return just the snippet content - no XML and no heading."
+
         # Ask is a special case, where the text to process is the prompted question, not the selected text
         if prompt.startswith("ask"):
             text_to_process = prompt.removeprefix("ask")
@@ -162,10 +167,6 @@ def gpt_apply_prompt(
         elif prompt == "pass":
             return text_to_process
 
-        match modifier:
-            case "snip":
-                prompt += "\n\nPlease return the response as a textmate snippet for insertion into an editor with placeholders that the user should edit. Return just the snippet content - no XML and no heading."
-
         return gpt_query(prompt, text_to_process)
 
     def gpt_help():
diff --git a/GPT/gpt.talon b/GPT/gpt.talon
index c2b42a7d..00637ca2 100644
--- a/GPT/gpt.talon
+++ b/GPT/gpt.talon
@@ -10,21 +10,24 @@ model <user.modelPrompt> [{user.modelSource}] [{user.modelDestination}]:
     result = user.gpt_apply_prompt(modelPrompt, text)
     user.gpt_insert_response(result, modelDestination or "")
 
-# Modifies a model command to be inserted as a snippet instead of a standard paste
+# Select the last GPT response so you can edit it further
+model take response: user.gpt_select_last()
+
+# Modifies a model command to be inserted as a snippet for VSCode instead of a standard paste
 # Otherwise same grammar as standard `model` command
 model snip <user.modelPrompt> [{user.modelSource}] [{user.modelDestination}]:
     text = user.gpt_get_source_text(modelSource or "")
     result = user.gpt_apply_prompt(modelPrompt, text, "snip")
     user.gpt_insert_response(result, modelDestination or "", "snip")
 
+# Modifies a model comand to always insert with the text selected
+# Useful for chaining together prompts immediately after they return
+# Otherwise same grammar as standard `model` command
 model chain <user.modelPrompt> [{user.modelSource}] [{user.modelDestination}]:
     text = user.gpt_get_source_text(modelSource or "")
     result = user.gpt_apply_prompt(modelPrompt, text)
     user.gpt_insert_response(result, modelDestination or "", "chain")
 
-# Select the last GPT response so you can edit it further
-model take response: user.gpt_select_last()
-
 # Applies an arbitrary prompt from the clipboard to selected text and pastes the result.
 # Useful for applying complex/custom prompts that need to be drafted in a text editor.
 model apply [from] clip$: