JavaScript DOM XSS via fetch().json() → insertAdjacentHTML not detected by CodeQL

從 JavaScript security-extended 查詢套件開始，檢視 fetch().json() 回應和 insertAdjacentHTML 的建模方式。判斷此流程是否被刻意排除，或是需要對來源和接收點進行建模，接著根據提供的範例驗證結論，並確認是否預期會得到 XSS 結果。

Description of the false positive

We are trying to better understand the design decisions behind JavaScript XSS detection in CodeQL, specifically around taint sources involving network responses.

I have a piece of client-side JavaScript that builds HTML using insertAdjacentHTML with values coming from a fetch().json() response. From an application-security perspective, this is treated as a potential DOM XSS risk in our project, but CodeQL does not report it.

I’d like to confirm whether this behavior is by design, and if so, what the recommended way is to model this trust boundary.

Code samples or links to source code

            function loadMessageLogs(pageSize, continuationToken) {
                let url = '?handler=LoadMessageLogs&pageSize=' + pageSize;
                if (continuationToken)
                    url += '&continuationToken=' + encodeURIComponent(continuationToken);

                fetch(url)
                    .then(response => response.json())
                    .then(data => {
                        const tbody = document.querySelector("#tblMessageLogs tbody");
                        tbody.innerHTML = "";

                        if (!data.items || data.items.length === 0) {
                            messageLogs.hidden = true;
                            noMessageLogs.hidden = false;
                        }
                        else {

                            data.items.forEach(item => {
                                const link = `<a href="/MessageLogs/Details/${item.messageId}">View message</a>`;
                                const row = `<tr>
                                        <td>${item.createdDate}</td>
                                        <td>${item.messageId}</td>
                                        <td>${item.interfaceId ?? ''}</td>
                                        <td>${item.target ?? ''}</td>
                                        <td>${item.mpanCore ?? ''}</td>
                                        <td>${item.meterId ?? ''}</td>
                                        <td>${link}</td>
                                        </tr>`;
                                tbody.insertAdjacentHTML('beforeend', row);
                            });

                            PagingModule.updatePaging(data.continuationToken)
                            messageLogs.hidden = false;
                            noMessageLogs.hidden = true;
                        }
                    });
            }

In our case, data.items[*] ultimately contains data that may originate from user input stored and returned by the backend.

• We are running the javascript-security-extended query suite.
• No XSS issue is reported for this code.