Skip to main content

What Pixee Fixes

This page is the canonical reference for what Pixee can triage and fix. It covers all vulnerability types and finding categories across SAST, SCA, and IaC findings.

Vulnerability Coverage Table

Vulnerability CategoryExamplesTriageFix Mode
Injection — SQLSQL injection via string concatenation, unparameterized queriesYesDeterministic / Both
Injection — CommandOS command injection, shell metacharacters in exec callsYesDeterministic / Both
Injection — LDAPLDAP injection via unsanitized directory search filtersYesAI
Injection — XPathXPath injection via user-controlled node selectionYesAI
Injection — NoSQLNoSQL injection in MongoDB, Redis, and similar query patternsYesAI
Injection — Expression LanguageEL/OGNL injection via template or framework evaluationYesAI
Cross-Site Scripting (XSS) — ReflectedReflected XSS via unencoded user input in HTTP responsesYesDeterministic / Both
Cross-Site Scripting (XSS) — StoredStored XSS persisted in database and rendered to other usersYesBoth
Cross-Site Scripting (XSS) — DOM-BasedDOM XSS via document.write, innerHTML, eval with user dataYesBoth
Path TraversalDirectory traversal via ../ sequences, file inclusion flawsYesDeterministic / Both
Server-Side Request Forgery (SSRF)SSRF via unvalidated URL parameters to internal or cloud metadata endpointsYesDeterministic / Both
Insecure DeserializationUnsafe deserialization of untrusted dataYesBoth
Weak CryptographyMD5, SHA-1 for security-sensitive operations; ECB mode; short key lengthsYesDeterministic / Both
Insecure RandomnessInsecure random number generators used for security-sensitive tokensYesDeterministic / Both
Hardcoded SecretsHardcoded API keys, passwords, tokens in source codeYesAI
Authentication / Session WeaknessesBroken authentication, session fixation, missing session invalidation, insecure cookie flagsYesBoth
Insecure Direct Object Reference (IDOR)Missing authorization checks on resource identifiersYesAI
XML / XXE VulnerabilitiesXML External Entity injection, DTD processing, XXE via SSRFYesDeterministic / Both
Insecure Temp FilesPredictable temp file creation, race conditions on tempfile usageYesDeterministic / Both
Logging Sensitive DataPII, credentials, tokens written to logsYesAI
Security MisconfigurationsMissing security headers, insecure defaults, disabled CSRF protectionYesBoth
SSL/TLS WeaknessesDisabled hostname verification, accepting all certificates, weak protocolsYesDeterministic / Both
Dependency Vulnerabilities (SCA)CVEs in direct and transitive open-source dependenciesYesDeterministic / Both
Container / Dockerfile MisconfigurationsRunning as root, insecure base image patterns, exposed secrets in layersYesBoth
Infrastructure as Code (IaC) MisconfigurationsTerraform / CloudFormation / Kubernetes / Helm security misconfigurations, overly permissive IAM, unencrypted storageYesBoth
Open RedirectRedirects to attacker-controlled URLs via unvalidated redirect parametersYesBoth
Code Injection / Eval InjectionDynamic code execution from untrusted inputYesBoth
Prototype PollutionJavaScript object prototype manipulation via untrusted key assignmentYesBoth
Template Injection (SSTI)Server-side template injection via user input rendered in template enginesYesBoth
Insecure File UploadUnrestricted file type or content upload without validationYesBoth
Missing Security HeadersCSP, HSTS, X-Frame-Options, and other protective headers absent from responsesYesBoth
CORS MisconfigurationOverly permissive cross-origin resource sharing allowing untrusted originsYesBoth
Race Conditions / TOCTOUTime-of-check to time-of-use vulnerabilities in file and resource accessYesTriage
Improper Input ValidationMissing or insufficient validation of input data before processingYesBoth
Integer Overflow / UnderflowArithmetic boundary condition errors leading to unexpected behaviorYesBoth

Secrets Detection

Pixee primarily triages secrets findings — detecting hardcoded credentials, API keys, tokens, and cloud provider secrets in source code. Automated fixes are available for common patterns where a safe remediation is unambiguous.

CategoryExamples
API Keys & TokensHardcoded API keys, OAuth tokens, service account keys
Credentials in CodeHardcoded passwords, database connection strings
Cloud Provider SecretsAWS access keys, GCP service account JSON, Azure credentials
Private Keys & CertificatesRSA/EC private keys, TLS certificates committed to repos

Custom Rules

Pixee's triage engine handles custom scanner rules — including custom Semgrep rules, custom CodeQL queries, and internal rule sets — through its adaptive analysis tier. Custom rules don't require Pixee configuration; the engine generates triage logic for novel rule types automatically.

Fix Modes Explained

Deterministic — A pre-built codemod applies a rule-based transformation. Same input always produces the same output. Zero LLM involvement. Fastest fix path and most predictable output.

AI — An AI-powered fix is generated for patterns where deterministic rules do not reach — custom framework wrappers, multi-file dataflows, context-dependent sanitization, or novel vulnerability patterns. Every AI-generated fix passes through an independent quality evaluation before delivery.

Both — A deterministic codemod handles well-understood patterns (standard libraries, known frameworks), while AI handles custom or complex variants. Routing is automatic.

Fix mode and coverage depth vary by language and scanner. For supported languages and IaC formats, see Language Support. For per-scanner setup, see Integrations.