Tool blast radius
Cap-jailed by default
Every tool is handed only the capabilities it needs — a scanner that got network
and one target can't wander your filesystem or creds. Isolation is the default, not a checkbox.
INDUSTRY PACKET · SECURITY RESEARCH & TOOLING
A research platform where every tool is jailed by default.
Red teams · Malware analysis · Reverse engineering · CTF & research
Security researchers run the most dangerous software on Earth — on the same box they keep their creds and client data on. sigilOS flips the default: the platform under your tools is memory-safe and verifiable, every tool runs capability-jailed, and an untrusted sample only reaches what you explicitly hand it.
The problem
The standard research distro is a pile of powerful tools on a general-purpose Linux with ambient authority — any tool, and anything it executes, runs with your reach. Analysts lean on full VMs for isolation, which are heavy, escapable, and easy to forget. And the OS itself is the same memory-unsafe C substrate that the samples you're studying are written to exploit. Isolation is opt-in, and one slip is a compromised host.
Why sigilOS fits
Untrusted samples
Quarantine + trust-on-diff
Drop a suspect binary or USB and it lands in a cap-isolated sandbox; it's trusted
only on a verified content-hash diff. Detonate what you want to observe, not your host.
The host itself
Memory-safe, verifiable base
The OS under your tools is all-Sigil and memory-safe — the substrate isn't the
same C attack surface the malware is hunting for.
Per-tab / per-target
EL0-isolated processes
Fault-isolated processes and a cap-native browser with per-tab isolation and a live
capability inspector — see exactly what a component can touch.
Remote & range work
SRDX secure lane
A kernel-native remote display/control transport for driving isolated range boxes
without a fragile RDP/VNC stack. SRDX →
Provenance
Signed, verified artifacts
The Relic-style container model — verify before you run — applies to tooling and
samples alike. Know an artifact is what it claims before it executes.
The pitch, in one line
Kali-style coverage, but every tool is capability-jailed and the platform underneath is memory-safe and verifiable — isolation by construction instead of by discipline. That's a research box you can actually trust to hold the line when a sample fights back.
Where we actually are — no hype
- The capability model, USB-quarantine trust, EL0 process isolation, and the cap-native browser with its capability inspector are real in-tree.
- IN PROGRESS The breadth of a mature research toolkit — the long tail of ported, cap-jailed tools — is early. We have the isolation model; the catalog is being built.
- HONEST If your workflow depends on a specific mature Linux toolchain today, we're not a drop-in replacement yet. If you want to help define what a secure-by-construction research OS looks like, this is the ground floor.
Who this is for
Red teamers, malware analysts, reverse engineers, and researchers who are done treating host isolation as a matter of remembering to spin up a VM — and who like the idea of an OS that assumes everything is hostile until it's explicitly trusted. Come break it; that's the point.