Version 1.2.2 is a methodological-integrity release. The THERP and CREAM human-reliability tools have both been rebuilt to follow their published canonical methods - THERP onto a true NUREG/CR-1278 event-tree engine, and CREAM onto Hollnagel's two-axis control-mode model - replacing earlier simplified calculations. A new Registers module introduces a project-wide reference layer for hazards, tasks, issues and documents; the SCTA editor is restructured for analyst clarity; and the Compliance module gains a periodic-review framework for operational programmes.
A new top-level module that centralises project-wide reference data in one place. The Registers module brings together four registers - Major Accident Hazards (MAH), a Task Register with built-in screening, an Issues Register, and a Document Registry - giving teams a single hub for the cross-cutting information that feeds their assessments. The MAH and Task registers support full CSV import and export, and the Task Register can classify tasks by risk level before an assessment method is selected.
The THERP tool has been rebuilt from the ground up to implement the published canonical NUREG/CR-1278 method. A true human-reliability event-tree engine replaces the previous flat-multiplier model, with the Chapter 10 dependence model (five discrete dependence levels and correct conditional-probability equations), discrete Table 20-16 performance-shaping factors keyed to task nature, the Chapter 12 diagnosis time-reliability model, and Error-Factor uncertainty bounds surfaced as a first-class result. Legacy flat-task THERP documents continue to open harmlessly alongside the new tree-based model.
The CREAM engine has been rearchitected to follow Hollnagel's canonical approach. Control-mode assessment now uses proper two-axis (reduced / improved) tallying with the published control-mode region boundaries, rather than a single scalar threshold. Dependency-adjustment rules are applied across the affected CPC pairs with their direct-effect preconditions enforced, and probability output is expressed as proper HEP intervals per control mode that reflect epistemic uncertainty.
The SCTA editor has been redesigned to reflect the way analysts actually work. The eight-step flow is now grouped into three clear phases - Study Setup, SCTA Study and Close-Out - with the three core study steps prominently numbered 1-2-3 throughout the navigator and progress tracking. The Human Error Analysis table moves into a dedicated section card with a live safety-critical step count and filters, and the error-prevention context is visually flagged as critical, all without changing the underlying step mechanics.
The Compliance module now supports both Acquisition and Operations lifecycle modes. A new periodic-review framework extends gate-review concepts into a recurring cycle with due, overdue and completed-this-cycle states and automatic next-due computation, while the health pill combines status and evidence-currency so stale evidence is flagged against its review interval. Operations mode surfaces a cadence-based review timeline that gives programme managers a clear view of review scheduling and currency across standards.
A method-fidelity audit found the previous THERP implementation departed from NUREG/CR-1278 in its calculation formula, its handling of task type, and the visibility of the Error Factor. All three findings are resolved by the new event-tree engine and domain model.
A method-fidelity audit found CREAM control-mode classification was using ad-hoc linear thresholds rather than Hollnagel's two-dimensional region boundaries. The scalar approach has been replaced with the correct two-axis tallying and canonical region lookup.
Register exports could previously corrupt any field containing line breaks - a common case for multi-line description or notes fields. A new shared CSV service uses RFC 4180 record-walking to preserve multi-line quoted fields correctly across the MAH, Task and Issues registers.
The Document Registry, previously buried as a sub-tab inside the PM reporting section, is now surfaced as a first-class page under the Registers module - clearer to find and easier to manage, while removing clutter from the reporting interface.
Every major method implementation is now audited for fidelity against its published reference - NUREG/CR-1278 for THERP and Hollnagel (1998) for CREAM. A running validation log documents any deliberate simplification or scope constraint in one place for external validators, and any departure from the canonical method is annotated inline with a citation to its authorising audit.
The rebuilt engines are backed by extensive automated test coverage - tree traversal, path classification and dependence handling for THERP, and effect mapping, dependency rules, control-mode regions and HEP-interval geometry for CREAM - providing a stable foundation for the calculators, editors and reports that build on them.
The Registers module ships complete and verified across every phase, from the module shell and hazard/task registers through task screening and CSV round-trips to the Document Registry extraction, with a full suite of tests covering dual-routing and CSV handling of embedded newlines and escaped quotes.