MandateOS is a mandate-bound judgment system. It produces auditable, append-only Judgment Records and may escalate to human authority; it never places trades.
The questions below address common governance, risk, and control concerns raised by non-engineers.
1) Is MandateOS a trading system?
No. MandateOS is a governance and judgment system that produces records describing an outcome: affirm alignment, recommend an exposure-level adjustment, or escalate. It does not route orders or place trades.
2) Does MandateOS place trades or mutate portfolios?
No. The system outputs read-only artifacts (Judgment Records and escalation artifacts) and does not mutate portfolio state. Any portfolio changes occur only through human-controlled processes outside this repository.
3) Does MandateOS make investment decisions?
MandateOS does not decide what to do with capital; it applies documented procedures under an explicit mandate to produce a judgment outcome and rationale. It may recommend exposure-level adjustments within authority bounds, or escalate when authority or confidence limits are reached. Final accountability and authorization remain with human fiduciaries.
4) Does the system learn over time?
Not autonomously. Institutional learning occurs through human review and versioned updates to mandates and procedures. The system does not perform trial-and-error learning or adapt its authority at runtime.
5) Is reinforcement learning used?
No. Reinforcement learning and reward optimization are outside the authority model. The system is designed for determinism, audit, and escalation discipline rather than self-optimization.
6) What triggers escalation?
Escalation is triggered when predefined authority boundaries are reached: hard constraint breaches, confidence falling below mandate minimums, conflicting objectives, or persistent uncertainty beyond procedural tolerance. Escalation is a designed outcome, not an error condition.
7) What happens after escalation?
The system produces an escalation artifact that routes the issue to human authority. Automated judgment does not proceed as if the decision were resolved; escalation indicates human review is required to authorize any next step.
8) Can mandates or procedures be changed through the UI?
No. The Judgment Viewer is a read-only, non-authoritative viewer of published artifacts. It does not provide controls to modify mandates, procedures, or records.
9) How is past behavior audited and reproduced?
Each Judgment Record binds to the mandate and procedure versions used at the time of judgment, and records are append-only for audit. Auditing is performed by reviewing the referenced authority sources and the emitted record. Reproducing exact outputs additionally depends on the runtime engine version used to run the procedure.
10) Who is accountable for outcomes?
Human fiduciaries remain accountable for mandates, procedure governance, and any actions taken in response to judgments or escalations. The system is designed to make authority explicit and decisions auditable, not to replace fiduciary responsibility.