Risk and Quality

matic treats risk and quality as operational primitives rather than afterthoughts. Every work item, task, milestone, and project carries a risk profile that determines guardrail strictness, HITL checkpoint placement, dry-run requirements, and budget gate enforcement, so governance constraints are derived from assessed risk instead of being applied uniformly. Quality is measured through structured evals, continuous telemetry, and outcome tracking, giving operators evidence-based visibility into agent performance and output integrity.

Risk Classes

Risk Classes explains the six operational categories used to classify registered risks across the org: budget, technical, dependency, governance, model, and external.

Risk Profiles

Risk Profiles describes how risk is characterized per entity, including failure impact, likelihood, required guardrails, HITL placement, dry-run flags, and regression indicators.

Risk Registry

Risk Registry covers the org-level aggregated view of all active risk profiles, including severity, ownership, mitigation status, and escalation state.

Quality Scoring

Quality Scoring shows how matic combines eval results, telemetry, and outcome data into actionable quality signals for operators and governance stakeholders.

Mitigation Strategies

Mitigation Strategies documents the concrete actions tied to risk profiles that reduce failure likelihood or impact, from context rot curation to baseline regression controls.

Dry-Run Requirements

Dry-Run Requirements explains the side-effect-free execution mode used to validate consequential work before committing live, including when it is required, how it is scoped, and how results are reviewed.

Evals

Evals introduces the structured assessment framework covering automated, human-scored, outcome-driven, coverage, and regression evals, along with the schemas, runners, and rubrics that power them.