False Positive: BrokenCryptoAlgorithm.ql reports nearby string literals even when the cipher call itself stays on AES.
Nobody has claimed this yet.
Assessment
- Difficulty
- 3/5
- Estimated time
- 1-2 days
- Newbie friendliness
- 52/100
Research direction
Start with Security/CWE/CWE-327/BrokenCryptoAlgorithm.ql and inspect how the query determines the transformation reaching Cipher.getInstance(...). Reproduce the reports with NegCase1_Var5.java, NegCase3.java, and NegCase4_Var5.java, then verify that safe AES calls are no longer reported while genuine insecure transformations remain detected.
Written by the indexing model from the issue text.
Description
Version
codeql 2.24.3
Checker
- Checker id:
Security/CWE/CWE-327/BrokenCryptoAlgorithm.ql - Checker description: This checker detects insecure cryptographic algorithms (e.g., DES, RC4, ECB mode) being used as the algorithm specification in cryptographic operations.
Description of the false positive
These samples all keep the actual cryptographic operation on a safe algorithm. What changes is that insecure-looking strings appear somewhere nearby, or an intermediate computation happens to mention part of an insecure transformation name.
That should not be enough to report Security/CWE/CWE-327/BrokenCryptoAlgorithm.ql. The rule should care about the transformation that reaches Cipher.getInstance(...), not unrelated literals in the same method.
Affected test cases
NegCase1_Var5.java
This case requests AES/CBC/PKCS5Padding. There is no insecure algorithm at the sink, so this should stay unreported.
// A program with no string literal matching the insecure cryptographic algorithm pattern should not be flagged as insecure crypto usage.
package scensct.var.neg;
public class NegCase1_Var5 {
// Variant 5: Lexical and idiomatic - different secure algorithm from same family
public static void main(String[] args) throws Exception {
// Using another secure algorithm: AES/CBC/PKCS5Padding (still secure if used with proper IV)
// This tests that the checker only flags insecure ones, not all algorithms.
String secureAlg = "AES/CBC/PKCS5Padding";
javax.crypto.Cipher cipher = javax.crypto.Cipher.getInstance(secureAlg); // [REPORTED LINE]
System.out.println("Cipher created with: " + secureAlg);
}
}
NegCase3.java
This sample keeps an RC4-looking string around, but the cipher call still uses AES/CBC/PKCS5Padding. Reporting this would mean the query is matching on lexical proximity rather than the actual call argument.
// A string literal matching the insecure pattern and a cryptographic call exist, but the call uses a different secure algorithm string should not be flagged as insecure crypto usage.
package scensct.core.neg;
public class NegCase3 {
public static void main(String[] args) throws Exception {
String unusedLiteral = "RC4_NOT_USED"; // Changed value to avoid matching insecure pattern
String secureAlg = "AES/CBC/PKCS5Padding"; // Different secure algorithm
javax.crypto.Cipher cipher = javax.crypto.Cipher.getInstance(secureAlg); // Crypto call uses secure string // [REPORTED LINE]
System.out.println("Insecure literal unused: " + unusedLiteral);
}
}
NegCase4_Var5.java
Here the code derives "DES" from a string, but that value never becomes the transformation passed to Cipher.getInstance(...). The live sink still uses AES, so this is not a real broken-crypto finding.
// A string literal matching the insecure pattern flows along a path but does not reach the algorithm parameter of a cryptographic operation should not be flagged as insecure crypto usage.
package scensct.var.neg;
public class NegCase4_Var5 {
public static void main(String[] args) throws Exception {
String alg = "DES/ECB/PKCS5Padding";
String extracted = null;
try {
extracted = alg.toUpperCase().substring(0, 3);
} catch (RuntimeException e) {
// ignore, won't happen
}
javax.crypto.Cipher cipher = javax.crypto.Cipher.getInstance("AES/CBC/PKCS5Padding"); // [REPORTED LINE]
System.out.println("Extracted: " + extracted);
}
}
Cause analysis
The common failure mode is that the query seems to treat suspicious string constants as if they were equivalent to the transformation that reaches the crypto API. That is too loose for Security/CWE/CWE-327/BrokenCryptoAlgorithm.ql.
In all three samples, the value at Cipher.getInstance(...) is still a safe AES transformation. The reported issue only appears if the query keeps too much context from nearby literals or partial string computations instead of following the exact argument that reaches the sink.
- Dominant language
- CodeQL
- Stars
- 10.1k
- Forks
- 2.1k
- Avg merge
- 2d 16h
- Merged PRs (30d)
- 143
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
More from github/codeql
-
agentic-workflows
Difficulty 2/5 1-3 hours Newbie friendliness 70/100
-
false-positive javascript
Difficulty 2/5 1-3 hours Newbie friendliness 84/100
-
C#: cs/simplifiable-boolean-expression false positive on Nullable<bool> compared with a literal Open
Difficulty 2/5 1-3 hours Newbie friendliness 82/100
-
Difficulty 2/5 1-3 hours Newbie friendliness 78/100
-
false-positive
Difficulty 2/5 1-3 hours Newbie friendliness 70/100
Similar issues
-
Difficulty 2/5 1-3 hours Newbie friendliness 78/100
punkpeye/mcp-remote#369 ·
-
Mend: dependency security vulnerability untriaged
Difficulty 1/5 Under an hour Newbie friendliness 86/100
-
Difficulty 2/5 1-3 hours Newbie friendliness 68/100
-
bug
Difficulty 1/5 Under an hour Newbie friendliness 90/100
cisagov/vulnrichment#337 ·
-
bug DUP Reservations
Difficulty 2/5 1-3 hours Newbie friendliness 68/100
bcgov/reserve-rec-public#896 ·