Agent skill · security · aradotso
vulnhunter-security-scanner
Agentic AI security tool that applies attacker-first analysis to source code, using falsification-based reasoning to minimize false positives
Why this skill is useful
Adds a structured workflow for vulnerability scanning and fixing, including executable commands and a falsification engine that enhances security analysis capabilities.
What it needs
Requires git, python installed locally. Requires claude account access. About 7k tokens when loaded. Last updated 2026-08-04. 11 stars on the source repository.
What this skill does
VulnHunter Security Scanner Skill by ara.so — Security Skills collection. VulnHunter is an agentic AI security tool that applies proactive, attacker-first analysis directly to source code. Unlike traditional SAST scanners that flag patterns and generate false positives, VulnHunter reasons like an adversary: it identifies which defects are actually exploitable, maps attack paths, and proposes evidence-backed fixes. Core Concepts Attacker-First Forward Analysis: Starts at attacker-accessible entry points (APIs, file uploads, network messages) and reasons forward to evaluate whether an attacker can truly break through, rather than working backward from dangerous sinks. Falsification Engine: After finding a potential vulnerability, VulnHunter runs a structured workflow to disprove its own argument, searching for flawed assumptions, logic gaps, or security controls that would block the attack. Three-Skill Loop: /vulnhunt - Hunt for vulnerabilities /vulnhunter-fix - Fix verified issues with test-driven remediation /vulnhunt-fix-verify - Independently verify fixes Prerequisites Claude Code CLI authenticated with access to Claude Opus (required) Python 3.12+ (for agent runtime and harness tooling) Git and GitHub CLI (gh) authenticated (for fix workflow) Anthropic Cyber Verification Program enrollment recommended to avoid cyber safeguard blocks Installation Note: install.sh copies files (not symlinks) because symlinks can break subagent functionality. Re-run after pulling updates. Core Usage 1. Running the Scanner (/vulnhunt) The scanner will: 1. Recon Phase: Map entry points and dangerous sinks 2. Parallel Hunt: Trace paths from entries to sinks 3. Adversarial Disprove: Apply falsification to eliminate false positives 4. Capability Filter: Verify actual exploitability 5. Output: Emit only verified issues with exploit paths and proposed fixes 2. …
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills aradotso/vulnhunter-security-scanner