Templates and predicates

This section provides an introduction to the Snyk Code Query Language standard library of templates and predicates, to give some practical examples of each predicate and template.

Methods, literals, and arguments

A basic capability of Snyk Code is to find method calls and reason about their arguments. The goal here is to discover certain patterns of method calls and their arguments and to check if certain properties hold for these objects.

Consider the following Python program to be analyzed and searched. If a similar program is provided, the same examples will work for any other language Snyk Code supports. The code does not need to be compiled to be queried.

def safesend(x, y):

def finalsend(x):

o = connect()
o1 = connect()
safesend(o, 'connect')
safesend(o, 'message')

The first query finds the method connect. The query "connect"returns both the string with this value and the method call with this name. These can be separated by putting the value in a template. We can use autocompletion to find how the name connect can be wrapped. Literal or StringLiteral will restrict the search results to the string 'connect', whereas CallExpression will restrict the results to the function call connect().

Note that you can find function calls to functions outside of the file being scanned. Trying to find CallExpression<"safesend"> will not yield results. The reason is that the analysis may inline local functions in order to reason about their behavior.

Look at the most called method in the file, send. This method is called on an object returned by connect and takes various strings as input. To see its arguments, you can use some of the templates for its arguments. These are HasArg0, HasArg1, and so on.

For example, you can find all calls to send on an object returned by connect with the following query:

CallExpression<"send"> HasArg0<CallExpression<"connect">> These are all locations, but we can find the places where we call send with the first argument taking the value connect.

This gives a different picture. The interprocedural analysis figured out that the message connect was sent in a call to a local function.

Different tasks around the state of the object o can be explored. Assume we want to find all the calls to send after disconnect. These should be pretty bad cases of the programs where the connection may be in a bad state. To do this query, you can perform the following query: CallExpression<"send"> HasArg0<DataFlowAfter<Arg0In<CallExpression<"disconnect">>>>

This query searches for calls to send with its argument 0 satisfying the following property: in the dataflow, it is after a location that is an argument 0 in a call to disconnect. This matches only the final unsafe send call.

For negation, you can search for objects that are returned by connect, but not calling disconnect for the returned object.

CallExpression<"connect"> Not<ForSameObject<Arg0In<CallExpression<"disconnect">>>>

Similarly, you can call send with connect but not call disconnect. The following example has no matches in the preceding code snippet:

CallExpression<"send"> HasArg0<CallExpression<"connect">> Not<HasArg0<ForSameObject<Arg0In<CallExpression<"disconnect">>>>>

In all cases, the auto-completion for the rules should guide the search through the examples to make it easier to write such queries and those that are even more complex.

Taint flows and data sources

In many cases, you want to ensure that certain types of data have no way to flow to certain sensitive locations in the program. This is often done for security reasons, both to ensure compliance and correctness.

The first important element to query is sensitive data sources. Snyk has built in the following set of hierarchical data sources that you can query:


The first category of sources (SourceServer) is defined for programs that implement servers. These sources are typically fully user-controllable. This means that a malicious user can use them to launch an attack against the application or that one needs to handle such data with additional care. For example, you may want to check that authentication is always performed or that some other property is enforced.

The non-server predicates also apply to programs that do not implement server functionality.

Each of the predicates in the SourceServer category is returned by querying PRED:SourceServer or PRED:AnySource. Consider the following TypeScript code example:

import { Request, Response, NextFunction } from 'express';

module.exports = function productReviews () {
 return (req: Request, res: Response, next: NextFunction) => {
   let user = req.signedCookies;
   console.log('Some message ' + user);

This implements a request handler for an express server. In this case, the code reads the user cookie and logs it on the console. This might be a security vulnerability and a compliance problem for many applications. The first capability of Snyk Code is that it can discover these cookie locations, and you can connect them to check a lot of properties about them. In this case, running a PRED:SourceCookie query will find the first line of the request handler.

You can now verify that cookies are handled correctly by the code. For example, you can check that cookies do not end up logged anywhere. You can try to use data flow or ForSameObject. In this case, report if the cookie is logged as part of some other object, string concatenation, or other simple transformation.

To achieve this, there is a taint analysis done with the taint predicate. This takes the following shape: Taint< source, sanitizer, sink > .

Source is the source of sensitive data, sanitizer gives code patterns that would transform the data to be non-sensitive, and sink is the location where the sensitive data should reach a report to be made. The report is then made at the sink location.

Now, consider finding places where the user is logged in. You can then use the following query:

Taint<PRED:SourceCookie, PRED:None, CallExpression<"log">>

Of course, one may want to say that if a cookie is hashed using the function hash123, then it is safe to be logged. Then, the query would look like: Taint<PRED:SourceCookie, CallExpression<"hash123">, CallExpression<"log">>

Predefined sinks and sanitizers

Using the preceding taint template, you can start writing vulnerability detectors. However, Snyk Code provides predicates for various types of vulnerabilities. For example, if you want to detect SQL injection, this can be performed fully with the following query:

Taint<PRED:AnySource, PRED:SqliSanitizer, PRED:SqliSink>

Of course, this assumes that any of the sources in AnySource (see the hierarchy above) is one that a malicious actor may control. For example, not every application is set up in a way that users control environment variables or command line arguments. If you want to find only such SQL injections, you can run a query like: Taint<Or<PRED:AnySource, PRED:SourceResourceAccess>, PRED:SqliSanitizer, PRED:SqliSink>

In addition to SQL injection, Snyk Code can detect tens of other vulnerabilities and has corresponding predicates accessible from search and custom rules. The number of predicates is growing over time, and more rules are getting open to modifications.

Predefined templates and predicates



A "catchall" rule. Matches on anything.


Matches on a range of potential data sinks, including server responses, file systems, database writes, external APIs, logging mechanisms, and other forms of data export or display.


Matches on various types of potentially user controlled data sources, both servers (e.g., HTTP parameters/header/body, URLs, cookies, etc.) or indirect ones (e.g., database fields, local files, I/O or environment variables).


Matches on potential XSS sources (Language support: Apex).


Matches on cleartext cookie storage sanitizers (Supported languages: C#).


Matches on cleartext cookie storage sinks (Supported languages: C#).


Matches on cleartext transmission sanitizers (Supported languages: C#).


Matches on cleartext transmission sinks (Supported languages: C#).


Matches on client XSS (e.g., DOMXSS) sanitizers (Supported languages: Javascript).


Matches on client XSS (e.g., DOMXSS) sinks (Supported languages: Javascript).


Matches on code injection sanitizers (Supported languages: C#, Java, JS, Kotlin, PHP, Python, Ruby, Scala, Swift, VB.NET).


Matches on code injection sinks (Supported languages: C#, Java, JS, Kotlin, PHP, Python, Ruby, Scala, Swift, VB.NET).


Matches on command injection sanitizers.


Matches on command injection sinks.


Matches on deserialization sanitizers.


Matches on deserialization sinks.


Matches on email content injection sanitizers (Language support: Apex, Go).


Matches on email content injection sinks (Language support: Apex, Go).


Matches on error message outputs (e.g., stacktraces) (Language support: C#, Go, Java, Javascript, Kotlin, PHP, Scala).


Matches on error message output sanitizers (Supported languages: C#, Java, Javascript, Kotlin, PHP, Scala).


Matches on error message output sinks (Supported languages: C#, Go, Java, Javascript, Kotlin, PHP, Scala).


Matches on file inclusion sanitizers (Supported languages: PHP).


Matches on file inclusion sinks (Supported languages: PHP).


Matches on information disclosure sanitizers (Supported languages: C#, Go, PHP).


Matches on information disclosure sinks (Supported languages: C#, Go, PHP).


Used to document which language variants are missing an implementation for various stdlib definitions


Matches on JNDI injection sanitizers (Supported languages: Java, Kotlin, Scala).


Matches on JNDI injection sinks (Supported languages: Java, Kotlin, Scala).


Matches on LDAP injection sanitizers (Supported languages: C++, C#, Java, Kotlin, Scala).


Matches on LDAP injection sinks (Supported languages: C++, C#, Java, Kotlin, Scala).


Matches on log-forging sanitizers (Supported languages: C#).


Matches on log-forging sinks (Supported languages: C#).


Matches on prototype memory corruption sanitizers (Supported languages: Swift).


Matches on NoSQL sanitizers (Supported languages: Java, Javascript, Python).


Matches on NoSQL sinks (Supported languages: Java, Javascript, Python).


An "anti-catchall" rule. Matches on nothing.


Matches on open-redirect sanitizers (Supported languages: Apex, C#, Go, Java, Javascript, Kotlin, PHP, Python, Scala, VB.NET).


Matches on open-redirect sinks (Supported languages: Apex, C#, Go, Java, Javascript, Kotlin, PHP, Python, Scala, VB.NET).


Matches on prototype memory operation sinks (Supported languages: Swift).


Matches on potential XSS sinks (Language support: Apex).


Matches on prototype pollution assignment sanitizers (Supported languages: Javascript).


Matches on prototype pollution assignment sinks (Supported languages: Javascript).


Matches on path-traversal sanitizers.


Matches on path-traversal sinks.


Matches on regular-expression denial-of-service sanitizers.


Matches on regular-expression denial-of-service sinks.


Matches on reflection sanitizers (Supported languages: Java, Ruby).


Matches on reflection sinks (Supported languages: Java, Ruby).


Matches on soqli sanitizers (Language support: Apex).


Matches on soqli sinks (Language support: Apex).


Matches on sosli sanitizers (Language support: Apex).


Matches on sosli sinks (Language support: Apex).


Matches on reading values that are coming from zip, tar or other archives.


Matches on reading command line arguments.


Matches on reading values that are coming from a client-side framework such as Android, SwiftUI, UIKit, the DOM of an HTML page.


Matches on reading sensitive data (Language support: C#, Go, PHP).


Matches on reading values of cookies in an http server. These values are of security interest, because they can be fully controlled by malicious users.


Matches on reading values that are coming from a database.


Matches on reading environment variables of a process.


Matches on reading values that are coming from files.


Matches on reading http request body in an http server. These values are of security interest, because they may be fully controlled by malicious actors.


Matches on the name and content of file uploaded to an http server. These values are of security interest, because they may be fully controlled by malicious actors.


Matches on reading values of http headers in a server. These values are of security interest, because they may be fully controlled by malicious actors.


Matches on reading values of http parameters in an http server. These values are of security interest, because they may be fully controlled by malicious actors.


Matches on reading values from the local environment of the running process. This includes command line arguments, standard input or environment variables.


Matches on reading values that are coming from a remote resource through network requests.


Matches on reading values that may be controlled by an adversary, but not directly by sending requests to a server. E.g. if an application fetches a value from a URL, an adversary in control of that URL may use it to control its content.


Matches on reading request URLs in a server. The URLs are of security interest, because they may be fully controlled by malicious actors.