The boring infrastructure
behind a smarter scan.
Xseth is an early-stage research project exploring how far LLM-driven orchestration can take routine penetration testing. We pair classic ProjectDiscovery tooling — Subfinder, Naabu, HTTPX, Nmap — with security-tuned language models to produce actionable risk reports faster than a human operator would.
Every step is driven by Redis queues, so the pipeline survives crashes, restarts, and partial outages. The LLM only sees structured profiles; the scan tools only see validated targets. The result is a system that feels less like a chatbot wrapped around nmap and more like an actual junior pentester you can hand a domain to.
This site is the marketing surface. The operator console lives at app.xseth.com — that's where the actual penetration testing happens.
/ principle 01
Safety first
SmartGate refuses hospitals, governments, schools, and other off-limits targets before a single packet leaves the box.
/ principle 02
Boring infrastructure
Postgres, Redis, Docker — nothing exotic. We’d rather ship today than spend a quarter writing a custom queue.
/ principle 03
Reports that matter
The output is a prioritized list of findings with evidence — not a raw .txt dump from each scanner.
// who is building it
The founders
Xseth is a two-person founding team — one building the engine, one carrying it to the people who need it.
Oussema Taleb
CTO & Co-Founder
Builds the engine. Background in distributed and low-level systems — the queue-driven pipeline, the scanner orchestration, and the cognitive core that reasons over a live world-model of each target.
LinkedIn
Syef Sid Ali Benkrief
CEO & Co-Founder
Carries the work to the people who need it. Background in security sales — shaping how Xseth reaches lean security teams, and keeping the product honest about what it does and what it does not.
LinkedIn