Protego field desk

Read the code. Review the risk.

Code Security Review

Paste JavaScript, TypeScript, Python, Go, Java or PHP to find security patterns worth reviewing, with line numbers and practical fixes. Free, no sign-in or daily quota.

Code never leaves your device. Analysis runs entirely in your browser.

Quick answer

Code Security Review checks pasted JavaScript, TypeScript, Python, Go, Java and PHP for risky patterns such as embedded credentials, injection sinks and disabled security checks. Analysis runs entirely in your browser; code never leaves your device. Review each finding in context, apply a suitable fix, and use a full SAST tool for deeper analysis.

Load a vulnerable sample

Up to 200 KB UTF-8. Code is treated as text and is never executed. Finding line buttons select the matching source line.

0 / 204,800 bytes1 source lines

Ready for a local code review.

What these checks can tell you

A match points to a risky API or configuration, not a confirmed vulnerability. Check whether data is untrusted, whether a guard already exists, and whether the suggested fix fits your library. No matches is not a security verdict. Use Semgrep or CodeQL and human review for deeper analysis.

  • Possible hardcoded credentialCWE-798
  • Dynamically constructed SQLCWE-89
  • Shell command execution needs reviewCWE-78
  • Dynamic code executionCWE-95
  • Potentially unsafe deserializationCWE-502
  • MD5 or SHA-1 near password handlingCWE-327
  • Legacy cipher API without an explicit IVCWE-327
  • ECB encryption modeCWE-327
  • Non-cryptographic randomness near a secretCWE-330
  • HTML injection sinkCWE-79
  • TLS certificate verification disabledCWE-295
  • Flask debug mode enabledCWE-489
  • JWT signature verification bypassCWE-347
  • Request input in a filesystem pathCWE-22
  • Request-controlled redirect destinationCWE-601
  • Wildcard CORS origin with credentialsCWE-942

CWE references link to MITRE. OWASP mappings use the OWASP Top 10 2021 categories as contextual guidance.

Check the deployed website too

Source review and a website scan answer different questions. After reviewing your code, use the existing free scanner to check your public website configuration.

Website scanner limits: guests 3/day, registered users 10/day. Pasted code is never passed to the scanner.

Free website vulnerability scan
  • TLS and certificates
  • Security headers
  • Cookie flags
  • SPF and DMARC
  • Exposed paths
  • WAF detection

Frequently asked questions

Does my code leave my device?

No. The analyzer runs in your browser without uploading code or calling an AI service. Code and findings stay in page memory until you clear or leave the page. Copy report writes a redacted report to your clipboard only when you request it. Session replay is disabled on this page; tool usage analytics contain only the tool name and selected language.

Which languages and security issues does it check?

Choose JavaScript, TypeScript, Python, Go, Java or PHP, or use the heuristic auto-detect option. Language-specific rules look for possible credentials, injection, unsafe deserialization, risky cryptography, HTML injection sinks, disabled TLS or JWT checks, Flask debug mode, path traversal hints, open redirects and wildcard CORS with credentials. Each result includes a line, CWE reference, contextual OWASP Top 10 2021 category and remediation example.

Does a clean result mean the code is secure?

No. This is a pattern-based review aid, not a substitute for SAST tools such as Semgrep or CodeQL, dependency scanning, testing or human review. It does not build an AST, resolve general imports or aliases, track data across files, or prove exploitability. Go math/rand imports are recognized to distinguish crypto/rand; some checks inspect one nearby assignment. False positives and false negatives are possible. Static HTML and parameterized queries are skipped where practical.

What are the input and coverage limits?

Paste up to 200 KB of UTF-8 code. The engine skips ordinary line and block comments and standalone Python docstrings. It treats strings as opaque tokens and does not inspect code inside template expressions. JavaScript regex literals, escapes, indirect flows and unusual syntax can confuse the lexer. Call and configuration inspection is bounded to 256 tokens, and output to 300 findings. A warning marks detected coverage limits. PHP heredoc/nowdoc, PHP closing tags and Java Unicode escapes trigger a coverage warning with snippets omitted because their text boundaries are not parsed. Go raw strings and Java text blocks are opaque. No account or daily quota applies to this client-only tool.

What are the Go, Java and PHP limitations?

These are basic API patterns. Go HTML template misuse and gob deserialization are not checked. Java checks cover direct or nearby JDBC query construction, recognizable ObjectInputStream receivers, empty trust checks, literal accepting hostname verifiers and request input passed to writer or .html sinks. JSP scriptlets, indirect writer flows, JWT and cookie flags are not checked. PHP XSS checks cover request values in echo/print with an exemption for enclosing HTML encoders, which does not prove safety in other output contexts. PHP string assertions and create_function are flagged as legacy migration concerns: string evaluation and create_function were removed in PHP 8. Indirect flows and custom validation require manual review.