vuln-seeker

CLI code security scanner — detects vulnerabilities with 25+ regex patterns and optional AI-powered triage, with JSON/Markdown reports and CI mode

Fast, regex-driven code security scanning with JSON/Markdown reports and optional AI-assisted triage.

Why

Security reviews are often delayed because lightweight checks are missing from day-to-day development. vuln-seeker is designed to fill that gap:

  • Quick local feedback before code reaches CI.
  • Broad language coverage via text pattern matching (not parser-dependent).
  • Actionable output with severity, location, rule, and remediation advice.
  • CI-friendly behavior with non-zero exits for policy enforcement.
  • Optional AI analysis to prioritize likely true positives and remediation order.

Installation

Requirements

  • Python 3.10+

Install from source

git clone https://github.com/vikasudasi/vuln-seeker
cd vuln-seeker
pip install -e .

Verify installation

vuln-seeker --version
vuln-seeker --help

Usage

Basic scan

vuln-seeker .

Scan a specific path with all patterns

vuln-seeker ./src --pattern-set all

CI mode (fail on high severity findings)

vuln-seeker . --ci

AI-assisted analysis

export VULN_SEEKER_API_KEY="your_api_key"
vuln-seeker . --ai gpt-4.1

CLI Reference

Option Type / Values Default Description
path positional path . Directory to scan recursively.
--ai MODEL string none Enables AI analysis and sets model name.
--ci flag off Exit with code 1 if HIGH findings are present.
--ignore PATTERN repeatable glob [] Ignore file/dir patterns.
--output json, markdown, both both Report format(s).
--pattern-set standard, extended, all standard Vulnerability rule coverage.

Detection Categories

  • SQL injection
  • XSS
  • Hardcoded secrets
  • Command injection
  • Path traversal
  • Insecure deserialization
  • Sensitive data exposure

Output Formats

Terminal: Color-coded findings with severity labels, file paths, descriptions, and fix suggestions.

JSON: Machine-readable report with tool, version, scanned_files, total_findings, severity_counts, and findings array.

Markdown: Human-readable report with scan metadata, severity summary, and detailed finding entries.

License

MIT