What problem does it solve? Reviewing changes to Mastra-based AI agents is risky because subtle defects in authorization, memory scoping, or workflow persistence can cause cross-tenant data leaks or unsafe side effects. This Skill provides a structured review checklist that catches material correctness and tenant-safety regressions before they ship. ## Core Features & Use Cases - Authority and Tenant-Safety Review: Verifies that server-verified user, tenant, and request context is preserved through agent, tool, and workflow execution, and that browser IDs or model output never establish authority. - Workflow Integrity Checks: Validates HITL approval binding, resume/callback rejection of stale or foreign authority, durable workflow persistence, and stop/abort behavior for protected side effects. - Package API Verification: Requires claims about Mastra beta package behavior to be proven against installed source/types or official migration guidance rather than stale examples. - Use Case: When a pull request modifies an MDE Mastra agent's memory configuration, use this Skill to confirm the memory/thread/resource scope cannot bleed data across users or tenants, with a concrete tenant A to tenant B failure case for any defect found. ## Quick Start Review this pull request's changes to the Mastra agent and tool definitions for tenant-safety and correctness regressions using the mastra-review checklist.
