security-reviewer

To find exploitable vulnerabilities in a code diff, review changes touching authentication middleware, public endpoints, or permission checks. Trace untrusted user input to dangerous sinks like SQL queries or shell commands to identify injection vectors and provide concrete remediation steps.

Detecting authentication bypasses in new endpoints involves checking for missing authentication on state-changing operations and verifying ownership checks. Ensure user A cannot access user B's resources and prevent privilege escalation from regular users to admin roles.

Checking for insecure deserialization in user input handling involves tracing untrusted input passed to functions like pickle, Marshal, or unserialize. This identifies potential remote code execution or object injection vulnerabilities in deserialized data payloads.

Yes, you can scan code diffs for hardcoded secrets and sensitive data in logs. Look for hardcoded API keys, tokens, or passwords in source files, and verify that credentials, PII, or session tokens are not written to logs, error messages, or URL parameters.

Yes, this security review covers SSRF and path traversal vulnerabilities. It identifies user-controlled URLs passed to server-side HTTP clients without allowlist validation and file paths reaching filesystem operations without canonicalization and boundary checks.

Threat modeling injection vectors across web services is limited to reviewing code diffs touching authentication middleware, public endpoints, and user input handling. It focuses on SQL, XSS, shell, and template injection but does not analyze non-web service architectures.