SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2609.03438 · September 2026

Do GUI Agents Know When Not to Act? Enabling Conflict-Aware Termination for Multimodal GUI Agents

Weiqiang Wang, Jun Lan, Huijia Zhu, Zhuosheng Zhang, Tianjie Ju, Chuanbiao Song, Pengzhou Cheng, Zheng Wu, Yansi Li, Zhaoyuan Huang

arXiv · PDF · Open in the Atlas

Code that ran

6 of 6 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 2 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 4 failures of our instrument; 0 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2609.03438.json · badge: /api/ran/2609.03438.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
serein356/ConflictGuard canonical (classifier label)
classifier basis: census
6 ran · 0 no recorded run
FunctionStatusWhere it lives
calculate_f1_score Ran
ran, honoured the contract we drafted
serein356/ConflictGuard/src/conflictguard/eval/run_eval_steered.py
code served (permissive licence) · get_code("493f2f78623ee6ce")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned float, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# Analyze the function:
# - predicted_str: string parameter
# - ground_truth_str: string parameter
# The function processes these strings by removing brackets, converting to lowercase,
# splitting by whitespace, and computing F1 score based on set intersection.

ARGS = {
    "predicted_str": "the quick brown fox",
    "ground_truth_str": "the quick fox jumps"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
compute_directions Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
serein356/ConflictGuard/src/conflictguard/vectors/extract.py
code served (permissive licence) · get_code("b3c44660ecd7d762")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np
from sklearn.decomposition import PCA
from tqdm import tqdm

# Set random seed for reproducibility
np.random.seed(0)

# Create layer_hiddens: dict mapping layer_id to numpy arrays of hidden states
# Each array should have shape (num_samples, hidden_dim)
# The function expects pairs of samples (positive/negative), so use even number of samples
# Using 6 samples (3 pairs) and hidden_dim=4

layer_hiddens = {
    0: np.random.randn(6, 4).astype(np.float32),
    1: np.random.randn(6, 4).astype(np.float32),
    2: np.random.randn(6, 4).astype(np.float32),
}

# Create layer_ids: iterable of layer indices to process
layer_ids = [0, 1, 2]

# Create method: one of the three valid options
method = 'pca_pairwise'

# Construct ARGS dictionary
ARGS = {
    'layer_hiddens': layer_hiddens,
    'layer_ids': layer_ids,
    'method': method,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn, tqdm — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
infer_swipe_direction Ran
ran; our contract draft was wrong, not the code
serein356/ConflictGuard/src/conflictguard/eval/run_eval_steered.py
code served (permissive licence) · get_code("4b3d38c0b4dbeb2e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

# Test case 1: Basic swipe detection using coordinate and coordinate2
ARGS = {
    'args': {
        'coordinate': [10, 20],
        'coordinate2': [50, 25]
    }
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
normalize_action_type Ran
ran; our contract draft was wrong, not the code
serein356/ConflictGuard/src/conflictguard/eval/run_eval_steered.py
code served (permissive licence) · get_code("6f0fb55fcc4d246d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# The function takes a single parameter: act_type (a string)
# It performs string operations: strip(), lower(), and membership checks
# No complex objects needed, just a string argument

ARGS = {
    'act_type': 'TERMINATE'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
normalize_image_path Ran
ran; our contract draft was wrong, not the code
serein356/ConflictGuard/src/conflictguard/paths.py
code served (permissive licence) · get_code("1e6cf5ee1212f46a")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import os
from pathlib import Path
import tempfile

# Create a temporary directory to use as a fake root
with tempfile.TemporaryDirectory() as tmpdir:
    # Set up environment variables for testing
    os.environ['CONFLICTGUI_ROOT'] = tmpdir
    os.environ['ANDROID_CONTROL_ROOT'] = tmpdir
    
    # Create some test paths
    test_dir = Path(tmpdir)
    (test_dir / 'data' / 'rawdata').mkdir(parents=True, exist_ok=True)
    (test_dir / 'CONFLICT' / 'subdir').mkdir(parents=True, exist_ok=True)
    (test_dir / 'android_control_parsed').mkdir(parents=True, exist_ok=True)
    
    ARGS = {
        'image_path': 'test_image.jpg'
    }
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
select_best_layer Ran
ran, honoured the contract we drafted
serein356/ConflictGuard/src/conflictguard/vectors/extract.py
code served (permissive licence) · get_code("a677e902304070cb")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned int, ndim 0, dtype family int, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

def select_best_layer(variances: dict) -> int:
    return max(variances, key=variances.__getitem__)

# Construct valid arguments
variances = {
    0: 0.15,
    1: 0.42,
    2: 0.28,
    3: 0.65,
    4: 0.31
}

ARGS = {
    "variances": variances
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type int, ndim 0, dtype family int, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block

Also present in other papers' code

Of 6 samples lifted from this paper's linked repositories, 6 carry a function identity and 0 do not. Of the 6 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 6 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Graphical user interface (GUI) agents are increasingly used to execute natural-language instructions on user interfaces, yet real users may issue infeasible instructions due to benign mistakes. A reliable agent should not only know how to act, but also when not to act. In this work, we introduce CONFLICTGUI, a benchmark covering instruction-internal conflicts and instruction-GUI context conflicts to study conflict-aware termination. Our evaluation reveals severe execution-biased overcompliance: agents that perform well on feasible tasks often continue to execute blindly under conflicting instructions. To mitigate this behavior, we propose CONFLICTGUARD, an inference-time framework that aligns an agent's feasibility awareness with its action generation. CONFLICTGUARD contains two coupled components: a feasibility verification protocol that guides the agent to assess instruction logic and GUI-side evidence before acting, and a conditional action modulation mechanism that steers agents from over-compliant execution into termination-oriented behavior. Experiments across five widely-used agents demonstrate that CONFLICTGUARD improves average conflict task success rate significantly, while preserving normal GUI-task performance. These results validate that a lightweight inference-time intervention can substantially boost GUI Agent's competence to identify inappropriate execution scenarios and refrain from unnecessary actions.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2609.03438")
get_code_for_paper("2609.03438")
have("2609.03438")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2609.03438.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2609.03438.svg)](https://syntology.ai/paper/2609.03438)

Connect an agent — have() is free.