invariant-policy-checker

name	invariant-policy-checker
description	Define and evaluate deterministic invariants and policy checks for DSDS simulations. Use when encoding correctness rules, post-run validation gates, compliance constraints, and safety checks that must hold across scenarios.

Invariant Policy Checker

Overview

Turn system correctness requirements into explicit simulation checks. Report violations with severity, evidence, and likely causes.

Workflow

1. Collect invariant candidates

• Identify invariants for idempotency, ordering, consistency, security, and SLOs.
• Convert vague requirements into measurable assertions.

2. Encode check definitions

• Define each check with id, condition, severity, and failure message.
• Map each condition to observable metrics or trace state.

3. Run post-simulation evaluation

• Evaluate checks against output metrics, traces, and events.
• Flag violations with context and impacted scope.

4. Classify and triage

• Separate hard failures from warning-level drift.
• Highlight violations that indicate systemic risk.

5. Recommend remediation

• Suggest configuration or architecture changes linked to each violation.
• Suggest follow-up scenarios to re-test fixes.

Output format

1. Invariant set
2. Violation report
3. Severity-ranked findings
4. Remediation plan
5. Follow-up validation scenarios

References

• Read references/invariant-catalog.md for invariant classes and examples.
• Use these repo sources:
• canonical-catalogue/DSDS Canonical Catalogue - Policies and invariants.csv
• planning/IMPLEMENTATION_PLAN.md (Phase 9.2)
• schema/complete_simulator_schema.ts (invariant types)