Whitepapers

How Scutum is built, deployed, secured, and scaled. For forward-looking work — open problems we're investigating but haven't shipped — see Research.

Operating Scutum

Self-Managed Best PracticesThe operator playbook: day-one setup, backups, monitoring, scaling, upgrades, disaster scenarios.

Gateway SecurityThreat model, trust boundaries, audit-log integrity, common attacks, rotation cadences. Used in customer security review.

Managed vs Self-HostedWhat you trade when you let us run it for you, what stays yours regardless. Migration paths in either direction.

How the platform works

Platform InternalsSRE agent in production, scaling patterns past a single host, semantic cache with PII boundaries. Three meaty capabilities, one paper.

The math behind specific subsystems

Risk-Bounded Autonomous RemediationHow the SRE agent decides whether to auto-execute or ask for approval. Four-component risk decomposition.

Cost-Aware Multi-Provider RoutingScalarised three-objective utility (cost / latency / quality) with operator-supplied weights in interpretable units.

---

A different way to slice the same six papers, by reader's role:

• You're an operator considering adopting Scutum: read Self-Managed Best Practices, then Managed vs Self-Hosted.
• You're a security reviewer / CISO: read Gateway Security.
• You're an engineer evaluating the platform's depth: read Platform Internals, then either of the two math papers depending on which subsystem you care about.
• You're a researcher curious about the formal framing: read Risk-Bounded Autonomous Remediation or Cost-Aware Multi-Provider Routing. Then look at the open problems in Research.

If a paper you wanted is missing, write to [email protected]. The set grows as we learn from operators.