LeakLens

Inspiration

With the rise of AI-assisted "vibe coding", more developers are rapidly shipping code without deeply reviewing security implications. In 2024, GitHub saw a 23.7M increase (~25% growth) in security risks, driven in part by faster, less-reviewed development workflows.

Existing secret scanners often require centralized code uploads, provide limited context, or only detect a small set of patterns. We built LeakLens, a privacy-first security assistant that scans code, explains risks, and helps developers fix vulnerabilities instantly.

What We Built

LeakLens is a local-first security scanner and auto-remediation platform that detects:

Exposed API keys, cloud credentials, and tokens Personally identifiable information (PII) Weak cryptography patterns Injection vulnerabilities Unsafe deserialization Insecure database/configuration patterns

The engine contains 102 security rules across 21 vulnerability categories, combining provider-specific detection patterns with intelligent heuristics.

Unlike traditional scanners that only report problems, LeakLens helps developers understand and fix them, most with a single click.

Each vulnerability finding provides:

Severity rating Confidence score Masked secret preview Explanation of impact Recommended remediation steps Technical Implementation

Built a lightweight Python HTTP server exposing APIs for:

File scanning Repository analysis AI review Automated remediation Fix instructions Detection Engine

The core scanner uses:

102 provider-tuned regex rules for known secret formats Entropy analysis using Shannon entropy to identify high-randomness secrets Context-aware heuristics to reduce false positives Severity/confidence scoring for prioritization Frontend

LeakLens goes beyond detection by safely fixing common vulnerabilities.

For supported hardcoded secrets, it:

Replaces credentials with environment variable references Creates .env.example placeholders Saves .bak backups for rollback Preserves manual review for sensitive cases like private keys

What's Next

Future improvements include:

AST-based language-aware vulnerability detection CI/CD pipeline integrations Expanded remediation capabilities Audit logs and rollback management Team-based security workflows