UEChecker: Detecting Unchecked External Call Vulnerabilities in DApps via Graph Analysis
Dechao Kong, Xiaoqi Li, Wenkai Li · 2025
Read UEChecker: Detecting Unchecked External Call Vulnerabilities in DApps via Graph Analysis by Dechao Kong, Xiaoqi Li, Wenkai Li (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-01343.txt.
Introduction
Dechao Kong, Xiaoqi Li, and Wenkai Li present UEChecker, a 2025 tool that hunts a common flaw in blockchain apps: smart contracts that call outside protocols without checking whether the call succeeded. Such unchecked calls open doors to flash loan and reentry attacks, and losses at the contract layer run to tens of billions of dollars.
The tool turns a contract's call pattern into a graph and studies it with a neural network built for graphs. Extra modules rebuild missing links, pool neighbor data, and track patterns at several scales at once. The team audited the contracts of 608 live apps to show the checker at work in the wild.
Catalog record
| CREATOR | Dechao Kong, Xiaoqi Li, Wenkai Li |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 33 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-01343
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
UEChecker: Detecting Unchecked External Call Vulnerabilities in DApps via Graph Analysis Dechao Kong, Xiaoqi Li, Wenkai Li Abstract The increasing number of attacks on the contract layer of DApps has resulted in economic losses amounting to $66 billion. Vulnerabilities arise when contracts interact with external protocols without verifying the results of the calls, leading to exploit entry points such as flash loan attacks and reentrancy attacks. In this paper, we propose UEChecker, a deep learning-based tool that utilizes a call graph and a Graph Convolutional Network to detect unchecked external call vulnerabilities. We design the following components: An edge prediction module that reconstructs the feature representation of nodes and edges in the call graph; A node aggregation module that captures structural information from both the node itself and its neighbors, thereby enhancing feature representation between nodes and improving the model’s understanding of the global graph structure; A Conformer Block module that integrates multi-head attention, convolutional modules, and feedforward neural networks to more effectively capture dependencies of different scales within the call graph, extending beyond immediate neighbors and enhancing the performance of vulnerability detection. Finally, we combine these modules with Graph Convolutional Network to detect unchecked external call vulnerabilities. By auditing the smart contracts of 608 DApps, our results show that our tool achieves an accuracy of 87.59% in detecting unchecked external call vulnerabilities. Furthermore, we compare our tool with GAT, LSTM, and GCN baselines, and in the comparison experiments, UEChecker consistently outperforms these models in terms of accuracy. 1 Introduction The total value locked in decentralized applications (DApps) and encrypted protocols has exceeded $4.4 billion, attracting more than 82 million active cryptocurrency users worldwide dappradar 2024 ; Niu et al. 2024 . By embedding established logic into smart contracts and storing them on the blockchain, these contracts can execute automatically without the need for centralized institutional intervention Li and others 2021 ; Li et al. 2017 . This emerging technology allows users to autonomously control their personal funds, benefiting from the security and transparency provided by smart contracts. As blockchain and cryptocurrencies mature, DApps are rapidly expanding into various sectors, including finance Angeris et al. 2021 , gaming, NFTs, and social media. However, the threats posed by smart contracts are also becoming increasingly severe, such as flash loan attacks Chen et al. 2024b and market manipulation. DApps demand extremely high security for smart contracts Xia et al. 2021 ; Duan et al. 2022 ; Li et al. 2024a . Traditional methods such as symbolic execution Luu et al. 2016 ; So et al. 2021 , static analysis Feist et al. 2019 ; Ghaleb et al. 2023 , and taint tracking Ghaleb et al. 2022 can quickly detect vulnerabilities but rely heavily on expert knowledge, often leading to false positives and consuming substantial computational resources Wang et al. 2021 ; Durieux et al. 2020 . In the context of DApps, where multiple contracts interoperate to achieve complex business logic, deep learning models have proven effective in identifying potential attack behaviors Gao 2020 ; Li et al. 2024e ; Li et al. 2024d ; Li et al. 2024b ; Liu and Li 2025 . However, detecting vulnerabilities in DApps still presents the following challenges. Dynamic invocation: DApps are composed of multiple contracts involving numerous modules and branches Li et al. 2024c ; Bu et al. 2025a ; Bu et al. 2025b . Different scenarios may arise when contracts are actually invoked, and this uncertainty in code logic execution paths makes it difficult to ensure the security of external contract calls. Node function diversification: Different nodes represent different functional roles, such as external function calls, external call checks, and event triggers. This requires extracting features of nodes and call edges from a functional perspective. To address these issues, we propose UEChecker, a novel DApp security auditing tool. This tool utilizes Surya to extract high-level semantic representations of call graphs from source code and then identifies comprehensive graph information through node and edge feature extraction, clarifying the call relationships between different nodes. The main contributions of this paper are as follows: • To the best of our knowledge, we are the first to propose UEChecker, utilizing source code function body information to construct control and data flow call graphs to detect unchecked external call vulnerabilities in DApps. UEChecker detects DApp vulnerabilities based on Graph Convolutional Networks. It converts source codes into call graphs leveraging Surya and then analyzes and learns the features of the call graph to detect unchecked external call vulnerabilities in DApps. • We validate the effectiveness of UEChecker in 608 DApps. Through auditing the vulnerability reports in DAppSCAN, we label contracts that explicitly involve unchecked external call vulnerabilities. We also conduct comparative analysis using baseline models and our proposed model, and experimental results demonstrate that our model achieves higher detection accuracy of 87.59%, effectively addressing the security issues of the complex vulnerabilities. 2 Motivation and Preliminary In this section, we primarily introduce the motivation behind our technology selection and the types of vulnerabilities that this paper focuses on Unchecked External Calls Vulnerabilities in DApp. 2.1 Model Selection Compared to other advanced models for detecting contract vulnerabilities, GCN demonstrates unique advantages in identifying unchecked external call vulnerabilities in DApps Li et al. 2024d . GCN effectively processes information from various graph data structures, particularly for Dapps involving multi-contract calls. Smart contracts can be generalized into graph structures, where functions or variables can be viewed as nodes, and the edges represent the calling relationships between functions or variables. Through convolution operations, GCN captures both local and global information of nodes in the graph domain. For GCN, a single node can be represented as an aggregation of the entire graph’s information, as shown in the formula H^{(l+1)}=\sigma\left(\tilde{A}H^{(l)}W^{(l)}\right) . This formula illustrates that each node in the graph is processed simultaneously, capturing more complex graph information features. Traditional models like RNNs and LSTMs require transforming graph-structured data into sequential data, necessitating the ordered aggregation of sequential information. This transformation often leads to the loss of graph data. The core of GCN lies in updating a node’s global representation through neighborhood aggregation, a mechanism that balances information from the node and its neighboring nodes, thereby capturing the graph’s contextual information in a higher dimension. This capability is crucial in handling unchecked external call vulnerabilities, as the occurrence of vulnerabilities relates to modifications of functions and variables across multiple contracts Chen et al. 2018 ; Zou et al. 2025 ; Liu et al. 2024 . 2.2 Vulnerability Characteristics Unchecked external call vulnerabilities typically occur when a smart contract neglects to check the return value after calling an external function, allowing malicious contracts to exploit the vulnerability Wu et al. 2025 ; Zhong et al. 2023 . Dapps consist of multiple contracts with complex calling chains, which can be intuitively represented as graphs. The connections between nodes are embedded into the feature representations. GCN can receive features from nodes and their neighbors through each layer’s convolution operation, ensuring that the contextual information of the vulnerability is effectively captured by GCN. Hierarchical information aggregation helps GCN extract features of unchecked external call vulnerabilities. For such vulnerabilities, GCN gradually aggregates and extracts all nodes’ features. By learning a weight matrix,