What if a security assessment engine could do more than discover vulnerabilities? What if it could question its findings, recover from execution failures, understand threat models, visualize attack paths, and follow a structured offensive security methodology?
That's exactly the direction I'm taking with HunterX.
I'm excited to share a major set of upgrades that brings HunterX closer to what I envision as a truly AI-directed autonomous security assessment engine.
This isn't simply about adding more tools or running more scans.
It's about making the entire assessment process smarter, more adaptive, more evidence-driven, and more reliable.
Here's what's new.
🔥 What's New in HunterX?
- Multi-Agent Consensus Reviewer Finding a potential vulnerability is one thing. Being confident that it's real is another.
I've introduced a configurable BaseReviewer interface, along with SimpleMockReviewer, to support evidence-chain review and improve finding confidence.
The objective is to reduce false positives and make evidence validation a more meaningful part of the assessment lifecycle.
- Self-Healing Debugger Tool failures shouldn't necessarily mean the end of an assessment.
With DebuggerAgent integrated into MissionExecutor, HunterX can now analyze tool execution errors, identify potential causes, and attempt retries using corrected parameters.
The goal is to make mission execution more resilient without compromising scope enforcement, authorization, or security policies.
An autonomous engine should know how to handle failure, not simply report it.
- Threat Modeling Integration Security testing should be driven by more than a list of discovered endpoints.
I've added:
ThreatModelParser
ThreatModelIntegrator
These components bring structured threat models, including STRIDE, into mission planning.
HunterX can use threat-model context to identify gaps between theoretical threats and the security tests actually performed.
This introduces a more structured, threat-driven approach to assessment coverage.
- Attack Path Visualization Individual findings don't always tell the complete story.
Sometimes, the real security impact becomes clear only when multiple weaknesses and relationships are connected.
The new AttackPathVisualizer, integrated into the correlation module, enables knowledge-graph exports in visualization-ready JSON formats.
This helps make attack paths, evidence relationships, and security analysis easier to understand and communicate.
- Offensive Security Doctrine Plugin This is one of the additions I'm particularly excited about.
I've introduced an internal plugin called:
skill_offensive_doctrine
Its purpose is to bring structured, professional offensive security methodology directly into HunterX.
It includes:
Dedicated HUNT MODE / DOCTRINE ON activation.
Live assessment context through hunterx state.
Specialized modes for RECON, WEB, API, LLM, MOBILE, SOURCE CODE, EXPLOITATION, and REPORTING.
TRIAGE and CHAIN workflows.
An ONLY QUESTION gate.
Engagement-type enforcement.
A Kill List and 17 Critical Rules.
Evidence Discipline, marker discipline, body-diff analysis, and statistical sampling.
Reconnaissance pipelines and canonical attack chains.
Web3 Immunefi assessment knowledge.
LLM agent injection taxonomy.
Security research patterns covering 2024–2026, including the SAML wave.
Impact-first offensive security writeups.
And importantly, the plugin follows a zero-break architecture: opt-in, isolated, and fully rollbackable.
It's designed to provide methodology and context without replacing HunterX's underlying execution and security controls.
🧪 Capability Verification Matrix
One thing I strongly believe in:
A feature shouldn't be considered complete simply because its code exists. It should be verifiable.
That's why I've also added a formal Capability Verification Matrix.
It covers:
Consensus Reviewer: reviewer disagreement and false-positive testing.
Self-Healing Debugger: failure recovery and bounded retries.
Threat Modeling: STRIDE coverage and gap analysis.
Attack Path Visualizer: graph integrity and evidence provenance.
Offensive Doctrine: activation, isolation, rollback, and policy enforcement.
The complete verification matrix is available at:
tests/capability_verification/capability_verification_matrix.md
This gives the new capabilities a dedicated verification structure rather than relying on feature descriptions alone.
📊 Reporting and Evidence
HunterX continues to support multiple reporting and evidence formats:
Markdown — Human-readable reports.
HTML — Visual security dashboards.
JSON — Machine-readable results.
SARIF 2.1 — Integration with VS Code and GitHub CodeQL workflows.
PDF — Professional document exports.
Evidence Package (ZIP) — Bundled assessment evidence.
The guiding principle remains the same:
Every finding should have traceability to observations, evidence, validation, proof, replay, impact, tool results, target intelligence, or explicit analyst reasoning.
🌍 HunterX Is Already Part of the Security Community
I'm also proud to see HunterX listed and featured in several well-known cybersecurity and open-source discovery resources:
OWASP Community – Vulnerability Scanning Tools: https://owasp.org/www-community/Vulnerability_Scanning_Tools
Awesome: https://awesome.re/
Awesome Red Teaming: https://github.com/0xMrNiko/Awesome-Red-Teaming
Awesome AI in Cybersecurity: https://github.com/ElNiak/awesome-ai-cybersecurity
These listings are an encouraging milestone for the project and a great motivation to keep pushing its development forward.
HunterX also supports Linux, macOS, and Windows.
🧠 The Bigger Picture
My vision for HunterX has never been to build another vulnerability scanner.
I want to build an engine that can:
Observe → Reason → Test → Validate → Prove → Correlate → Reassess → Report.
An engine where AI provides strategic direction, while the execution layer enforces scope, authorization, safety, and evidence integrity.
An engine that doesn't just tell you what might be vulnerable, but works toward establishing what is actually vulnerable, why it matters, and what evidence supports that conclusion.
There's still a lot to build, improve, and validate. That's part of the journey.
But every iteration brings HunterX closer to that vision.
🚀 Explore HunterX
Official Website: https://nullc0d30.github.io/HunterX/
GitHub Repository: https://github.com/nullc0d30/HunterX
PyPI: https://pypi.org/project/hunterxsec/
Docker: nullc0d30/hunterx - Docker Image
Created and maintained by Ahmed Awad (NullC0d3).
If you're working in Cybersecurity, Threat Intelligence, Penetration Testing, Red Teaming, or AI-driven Security Research, I'd genuinely love to hear your thoughts.
What do you think an autonomous security assessment engine should be capable of doing next?
