WikiFuzzer

Author’s pick · Best Project. The product I present first when talking about engineering work beyond research and open-source tools.

WikiFuzzer is a local / on-prem Agentic AI platform for authorized Web and API penetration testing. It analyzes the attack surface, chooses the next best action, runs controlled live tests, and promotes findings to Confirmed only when Oracle and Verifier evidence succeeds.

Why it matters

• Data stays inside the organization — model, tools, and evidence do not depend on a public cloud assistant.
• Propose ≠ Confirm — LLM suggestions never become Confirmed alone.
• Real execution — controlled packs, auth, and browser paths; not text-only advice.
• CPU-first design — no requirement for a powerful GPU for the core workflow.

Coverage today

• Web / API security assessment (available)
• Roadmap: Network, IoT/SCADA, Mobile, AI security assessment, MLSecOps tooling

Links

• Product site: wikifuzzer.mhsec.me
• Request an organizational demo from the product page