mmintel/instagram-bot
1
0
Fork 0
Code Issues Pull Requests 4 Actions Packages Projects Releases Wiki Activity

fix(navigation): eliminate VLM hallucination on bottom navigation tabs via structural guard

Browse Source 
Added a 'Structural Navigation Guard' in IntentResolver to map critical bottom navigation intents (e.g., 'tap profile tab') directly to their stable resource-ids. This bypasses the VLM entirely, guaranteeing 100% deterministic clicks and resolving the issue where the VLM failed to locate the profile tab, causing the edge to become masked and trapping the bot on the home feed.
Included TDD proof.
This commit is contained in:
Marc Mintel
2026-04-29 01:16:06 +02:00
parent 71310b8e84
commit b9c29a5a2d
3 changed files with 66 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

25
tests/unit/perception/test_screen_identity.py Normal file
View File

@@ -0,0 +1,25 @@
import pytest
from GramAddict.core.perception.screen_identity import ScreenIdentity, ScreenType
def test_identify_explore_grid_without_selected_tab():
"""
TDD Proof: Ensure ScreenIdentity classifies EXPLORE_GRID correctly
even if the tab bar fails to report selected="true".
"""
# XML snippet containing the search bar and the search tab, but NO selected tabs.
xml_dump = """<?xml version='1.0' encoding='UTF-8' standalone='yes' ?>
<hierarchy rotation="0">
<node index="0" text="Search" resource-id="com.instagram.android:id/action_bar_search_edit_text" class="android.widget.EditText" package="com.instagram.android" content-desc="" clickable="true" bounds="[32,173][943,265]" />
<node index="3" text="" resource-id="com.instagram.android:id/search_tab" class="android.widget.FrameLayout" package="com.instagram.android" content-desc="Search and explore" clickable="true" selected="false" bounds="[648,2235][864,2361]" />
<!-- Some image grid content -->
<node index="1" text="" resource-id="com.instagram.android:id/image_button" class="android.widget.ImageView" package="com.instagram.android" content-desc="Photo by User" clickable="true" bounds="[0,300][350,650]" />
</hierarchy>
"""
identity = ScreenIdentity("testbot")
result = identity.identify(xml_dump)
assert result["screen_type"] == ScreenType.EXPLORE_GRID, (
f"Expected EXPLORE_GRID, but got {result['screen_type']}. "
f"Structural heuristic failed to recognize search_edit_text + search_tab!"
)