• Fixed a bug, first introduced in 2.20.3, that would prevent v-html attributes in Vue files from being flagged by the js/xss query. The original behaviour has been restored and the v-html attribute is once again functioning as a sink for the js/xss query.
• Fixed a bug that would in rare cases cause some regexp-based checks to be seen as generic taint sanitisers, even though the underlying regexp is not restrictive enough. The regexps are now analysed more precisely, and unrestrictive regexp checks will no longer block taint flow.
• Fixed a recently-introduced bug that caused js/server-side-unvalidated-url-redirection to ignore valid hostname checks and report spurious alerts after such a check. The original behaviour has been restored.

No user-facing changes.

• Improved precision of data flow through arrays, fixing some spurious flows that would sometimes cause the length property of an array to be seen as tainted.
• Improved call resolution logic to better handle calls resolving "downwards", targeting a method declared in a subclass of the enclosing class. Data flow analysis has also improved to avoid spurious flow between unrelated classes in the class hierarchy.

• Fixed a recently-introduced bug that prevented taint tracking through URLSearchParams objects. The original behaviour has been restored and taint should once again be tracked through such objects.
• Fixed a rare issue that would occur when a function declaration inside a block statement was referenced before it was declared. Such code is reliant on legacy web semantics, which is non-standard but nevertheless implemented by most engines. CodeQL now takes legacy web semantics into account and resolves references to these functions correctly.
• Fixed a bug that would cause parse errors in .jsx files in rare cases where the file contained syntax that was misinterpreted as Flow syntax.

• Improved support for NestJS applications that make use of dependency injection with custom providers. Calls to methods on an injected service should now be resolved properly.
• TypeScript extraction is now better at analyzing projects where the main tsconfig.json file does not include any source files, but references other tsconfig.json-like files that do include source files.
• The js/incorrect-suffix-check query now recognises some good patterns of the form origin.indexOf("." + allowedOrigin) that were previously falsely flagged.
• Added a new threat model kind called view-component-input, which can enabled with advanced setup. When enabled, all React props, Vue props, and input fields in an Angular component are seen as taint sources, even if none of the corresponding instantiation sites appear to pass in a tainted value. Some users may prefer this as a "defense in depth" option but note that it may result in false positives. Regardless of whether the threat model is enabled, CodeQL will propagate taint from the instantiation sites of such components into the components themselves.

• Fixed a bug that would occur when TypeScript code was found in an HTML-like file, such as a .vue file, but where it could not be associated with any tsconfig.json file. Previously the embedded code was not extracted in this case, but should now be extracted properly.

• The js/xss-through-dom query now recognises sources of DOM input originating from Angular templates.

• Fixed a TypeScript extractor crash that would occur when encountering an export specifier whose local specifier was a string literal.

No user-facing changes.

No user-facing changes.

No user-facing changes.

No user-facing changes.

No user-facing changes.

No user-facing changes.

• Added a new query (js/actions/actions-artifact-leak) to detect GitHub Actions artifacts that may leak the GITHUB_TOKEN token.

No user-facing changes.

• Message events in the browser are now properly classified as client-side taint sources. Previously they were incorrectly classified as server-side taint sources, which resulted in some alerts being reported by the wrong query, such as server-side URL redirection instead of client-side URL redirection.

No user-facing changes.

• Added a new query, js/insecure-helmet-configuration, to detect instances where Helmet middleware is configured with important security features disabled.

• Added a new query, js/functionality-from-untrusted-domain, which detects uses in HTML and JavaScript scripts from untrusted domains, including the polyfill.io content delivery network
  • it can be extended to detect other compromised scripts using user-provided data extensions of the untrustedDomain predicate, which takes one string argument with the domain to warn on (and will warn on any subdomains too).
• Modified existing query, js/functionality-from-untrusted-source, to allow adding this new query, but reusing the same logic
  • Added the ability to use data extensions to require SRI on CDN hostnames using the isCdnDomainWithCheckingRequired predicate, which takes one string argument of the full hostname to require SRI for.
• Created a new library, semmle.javascript.security.FunctionalityFromUntrustedSource, to support both queries.

• Added a new experimental query, js/cors-misconfiguration, which detects misconfigured CORS HTTP headers in the cors and apollo libraries.

No user-facing changes.

No user-facing changes.

• CodeQL package management is now generally available, and all GitHub-produced CodeQL packages have had their version numbers increased to 1.0.0.

No user-facing changes.

• The JavaScript extractor will on longer report syntax errors related to "strict mode". Files containing such errors are now being fully analyzed along with other sources files. This improves our support for source files that technically break the "strict mode" rules, but where a build steps transforms the code such that it ends up working at runtime.

• API::Node#getInstance() now includes instances of subclasses, include transitive subclasses. The same changes applies to uses of the Instance token in data extensions.

• The @precision of the js/unsafe-external-link has been reduced to low to reflect the fact that modern browsers do not expose the opening window for such links. This mitigates the potential security risk of having a link with target="_blank".

• The call graph has been improved, leading to more alerts for data flow based queries.

No user-facing changes.

No user-facing changes.

No user-facing changes.

• The left operand of the && operator no longer propagates data flow by default.

No user-facing changes.

• Added support for doT templates.

No user-facing changes.

No user-facing changes.

• Added django URLs to detected "safe" URL patterns in js/unsafe-external-link.

• Lower the security severity of log-injection to medium.
• Increase the security severity of XSS to high.

• Added modeling for importing express-rate-limit using a named import.

• Added the AmdModuleDefinition::Range class, making it possible to define custom aliases for the AMD define function.

No user-facing changes.

• Fixed an extractor crash that could occur in projects containing TypeScript files larger than 10 MB.

• Files larger than 10 MB are no longer be extracted or analyzed.
• Imports can now be resolved in more cases, where a non-constant string expression is passed to a require() call.

• Fixed an extractor crash that would occur in rare cases when a TypeScript file contains a self-referential namespace alias.

No user-facing changes.

No user-facing changes.

• The fs/promises package is now recognised as an alias for require('fs').promises.
• The js/path-injection query can now track taint through calls to path.join() with a spread argument, such as path.join(baseDir, ...args).

• The query "Arbitrary file write during zip extraction ("Zip Slip")" (js/zipslip) has been renamed to "Arbitrary file access during archive extraction ("Zip Slip")."

No user-facing changes.

• Fixed an issue where calls to a method named search would lead to false positive alerts related to regular expressions. This happened when the call was incorrectly seen as a call to String.prototype.search, since this function converts its first argument to a regular expression. The analysis is now more restrictive about when to treat search calls as regular expression sinks.

• Added taint sources from the @actions/core and @actions/github packages.
• Added command-injection sinks from the @actions/exec package.

• The js/indirect-command-line-injection query no longer flags command arguments that cannot be interpreted as a shell string.
• The js/unsafe-deserialization query no longer flags deserialization through the js-yaml library, except when it is used with an unsafe schema.
• The Forge module in CryptoLibraries.qll now correctly classifies SHA-512/224, SHA-512/256, and SHA-512/384 hashes used in message digests as NonKeyCiphers.

• Fixed a spurious diagnostic warning about comments in JSON files being illegal. Comments in JSON files are in fact fully supported, and the diagnostic message was misleading.

• Improved the call graph to better handle the case where a function is stored on a plain object and subsequently copied to a new host object via an extend call.

• Fixes an issue that would cause TypeScript extraction to hang in rare cases when extracting code containing recursive generic type aliases.

• The DisablingCertificateValidation.ql query has been updated to check createServer from https for disabled certificate validation.
• Improved the model of jQuery to account for XSS sinks where the HTML string is provided via a callback. This may lead to more results for the js/xss query.
• The js/weak-cryptographic-algorithm query now flags cryptograhic operations using a weak block mode, such as AES-ECB.

• Fixed a bug where a destructuring pattern could not be parsed if it had a property named get or set with a default value.

No user-facing changes.

• The following queries now recognize HTML sanitizers as propagating taint: js/sql-injection, js/path-injection, js/server-side-unvalidated-url-redirection, js/client-side-unvalidated-url-redirection, and js/request-forgery.

• The js/regex-injection query now recognizes environment variables and command-line arguments as sources.

No user-facing changes.

No user-facing changes.

No user-facing changes.

• The AlertSuppression.ql query has been updated to support the new // codeql[query-id] supression comments. These comments can be used to suppress an alert and must be placed on a blank line before the alert. In addition the legacy // lgtm and // lgtm[query-id] comments can now also be placed on the line before an alert.

No user-facing changes.

No user-facing changes.

• Added support for @hapi/glue and Hapi plugins to the frameworks/Hapi.qll library.

• Fixed a bug that would cause the extractor to crash when an import type is used in the extends clause of an interface.
• Fixed an issue with multi-line strings in YAML files being associated with an invalid location, causing alerts related to such strings to appear at the top of the YAML file.

• Added a new query, js/second-order-command-line-injection, to detect shell commands that may execute arbitrary code when the user has control over the arguments to a command-line program. This currently flags up unsafe invocations of git and hg.

• Added sources for user defined path and query parameters in Next.js.
• The alert message of many queries have been changed to better follow the style guide and make the message consistent with other languages.

• Removed some false positives from the js/file-system-race query by requiring that the file-check dominates the file-access.
• Improved taint tracking through JSON.stringify in cases where a tainted value is stored somewhere in the input object.

No user-facing changes.

• Improved how the JavaScript parser handles ambiguities between plain JavaScript and dialects such as Flow and E4X that use the same file extension. The parser now prefers plain JavaScript if possible, falling back to dialects only if the source code can not be parsed as plain JavaScript. Previously, there were rare cases where parsing would fail because the parser would erroneously attempt to parse dialect-specific syntax in a regular JavaScript file.

• The js/regexp/always-matches query will no longer report an empty regular expression as always matching, as this is often the intended behavior.

• The alert message of many queries have been changed to make the message consistent with other languages.

• Fixed a bug in the js/type-confusion-through-parameter-tampering query that would cause it to ignore sanitizers in branching conditions. The query should now report fewer false positives.

• Added a new query, py/suspicious-regexp-range, to detect character ranges in regular expressions that seem to match too many characters.

• A new query "Case-sensitive middleware path" (js/case-sensitive-middleware-path) has been added. It highlights middleware routes that can be bypassed due to having a case-sensitive regular expression path.

• Contextual queries and the query libraries they depend on have been moved to the codeql/javascript-all package.

• The js/resource-exhaustion query no longer treats the 3-argument version of Buffer.from as a sink, since it does not allocate a new buffer.

• The js/actions/command-injection query has been added. It highlights GitHub Actions workflows that may allow an attacker to execute arbitrary code in the workflow. The query previously existed an experimental query.
• A new query js/insecure-temporary-file has been added. The query detects the creation of temporary files that may be accessible by others users. The query is not run by default.

• The js/missing-origin-check query has been added. It highlights "message" event handlers that do not check the origin of the event.
  The query previously existed as the experimental js/missing-postmessageorigin-verification query.

• The call graph now deals more precisely with calls to accessors (getters and setters). Previously, calls to static accessors were not resolved, and some method calls were incorrectly seen as calls to an accessor. Both issues have been fixed.

• The js/resource-exhaustion query has been added. It highlights locations where an attacker can cause a large amount of resources to be consumed. The query previously existed as an experimental query.

• Improved handling of custom DOM elements, potentially leading to more alerts for the XSS queries.
• Improved taint tracking through calls to the Array.prototype.reduce function.

• Fixed an issue that would sometimes prevent the data-flow analysis from finding flow paths through a function that stores its result on an object. This may lead to more results for the security queries.

• A new query, js/functionality-from-untrusted-source, has been added to the query suite. It finds DOM elements that load functionality from untrusted sources, like script or iframe elements using http links. The query is run by default.

• The js/request-forgery query previously flagged both server-side and client-side request forgery, but these are now handled by two different queries:
  • js/request-forgery is now specific to server-side request forgery. Its precision has been raised to high and is now shown by default (it was previously in the security-extended suite).
  • js/client-side-request-forgery is specific to client-side request forgery. This is technically a new query but simply flags a subset of what the old query did. This has precision medium and is part of the security-extended suite.

• Added dataflow through the snapdragon library.

• A new query, js/unsafe-code-construction, has been added to the query suite, highlighting libraries that may leave clients vulnerable to arbitrary code execution. The query is not run by default.
• A new query js/file-system-race has been added. The query detects when there is time between a file being checked and used. The query is not run by default.
• A new query js/jwt-missing-verification has been added. The query detects applications that don't verify JWT tokens.
• The js/insecure-dependency query has been added. It detects dependencies that are downloaded using an unencrypted connection.

• A new query js/samesite-none-cookie has been added. The query detects when the SameSite attribute is set to None on a sensitive cookie.
• A new query js/empty-password-in-configuration-file has been added. The query detects empty passwords in configuration files. The query is not run by default.

• Support for handlebars templates has improved. Raw interpolation tags of the form {{& ... }} are now recognized, as well as whitespace-trimming tags like {{~ ... }}.
• Data flow is now tracked across middleware functions in more cases, leading to more security results in general. Affected packages are express and fastify.
• js/missing-token-validation has been made more precise, yielding both fewer false positives and more true positives.

• TypeScript 4.5 is now supported.

• The js/sensitive-get-query query has been added. It highlights GET requests that read sensitive information from the query string.
• The js/insufficient-key-size query has been added. It highlights the creation of cryptographic keys with a short key size.
• The js/session-fixation query has been added. It highlights servers that reuse a session after a user has logged in.