Skip to content

Commit 135b3ea

Browse files
committed
New click prompt 'no button' approach for SYSTEM_PROMPT_OCR_MAC
1 parent ff35f17 commit 135b3ea

File tree

1 file changed

+4
-4
lines changed

1 file changed

+4
-4
lines changed

operate/models/prompts.py

Lines changed: 4 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -208,7 +208,7 @@
208208
You have 4 possible operation actions available to you. The `pyautogui` library will be used to execute your decision. Your output will be used in a `json.loads` loads statement.
209209
210210
1. click - Move mouse and click
211-
[{{ "thought": "write a thought here", "operation": "click", "text": "The text in the button or link to click" }}] # Look for buttons and links with the text. We've hooked up the `pyautogui` so that you can click on any buttons or links as long as you have the text for them.
211+
[{{ "thought": "write a thought here", "operation": "click", "text": "The text in the button or link to click" }}] # Look for buttons or links with text to click. If the button you want to click doesn't have text you can say `"no button"` for the text value and we'll try a different method.
212212
213213
2. write - Write with your keyboard
214214
[{{ "thought": "write a thought here", "operation": "write", "content": "text to write here" }}]
@@ -233,7 +233,7 @@
233233
# Go to a website (LinkedIn) when the browser is already open
234234
235235
[
236-
{{ "thought": "I can see that Google Chrome is open. I'll focus on the address bar to type ", "operation": "press", "keys": ["command", "l"] }},
236+
{{ "thought": "I can see that Google Chrome is open. I'll focus on the address bar to type ", "operation": "press", "keys": ["command", "t"] }},
237237
{{ "thought": "Now I'll write LinkedIn's website to go there", "operation": "write", "content": "https://www.linkedin.com/feed/" }},
238238
{{ "thought": "Finally I'll press enter to go to LinkedIn", "operation": "press", "keys": ["enter"] }}
239239
]
@@ -258,7 +258,7 @@
258258
You have 4 possible operation actions available to you. The `pyautogui` library will be used to execute your decision. Your output will be used in a `json.loads` loads statement.
259259
260260
1. click - Move mouse and click
261-
[{{ "thought": "write a thought here", "operation": "click", "text": "The text in the button or link to click" }}] # Look for buttons and links with the text. We've hooked up the `pyautogui` so that you can click on any buttons or links as long as you have the text for them.
261+
[{{ "thought": "write a thought here", "operation": "click", "text": "The text in the button or link to click" }}] # Look for buttons or links with text to click. If the button you want to click doesn't have text you can say `"no button"` for the text value and we'll try a different method.
262262
263263
2. write - Write with your keyboard
264264
[{{ "thought": "write a thought here", "operation": "write", "content": "text to write here" }}]
@@ -283,7 +283,7 @@
283283
# Go to a website (LinkedIn) when the browser is already open
284284
285285
[
286-
{{ "thought": "I can see that Google Chrome is open. I'll focus on the address bar to type ", "operation": "press", "keys": ["ctrl", "l"] }},
286+
{{ "thought": "I can see that Google Chrome is open. I'll focus on the address bar to type ", "operation": "press", "keys": ["ctrl", "t"] }},
287287
{{ "thought": "Now I'll write LinkedIn's website to go there", "operation": "write", "content": "https://www.linkedin.com/feed/" }},
288288
{{ "thought": "Finally I'll press enter to go to LinkedIn", "operation": "press", "keys": ["enter"] }}
289289
]

0 commit comments

Comments
 (0)