agent-delegation

Default engineering control for agent work.

Runs the closed loop from defect discovery through root cause, regression proof, staging, production verification, and recurrence prevention.

Turns an ambiguous engineering request into a bounded, testable task contract for humans or coding agents.

Reviews identity, authorization, tenant scope, service trust, webhooks, signatures, replay protection, and negative security behavior.

Verifies semantic and behavioral correctness of LLM, retrieval, routing, memory, fallback, and tool-using agent systems.

Verifies that the exact intended artifact, database state, dependencies, and user path are healthy before and after promotion.

Governs replicated entities, async consumers, retries, deletion propagation, dead letters, and reconciliation across service boundaries.

Production bundle combining baseline, issue, contract, deployment, data, configuration, observability, and security controls.

Detects and prevents drift between backend routes, schemas, serializers, clients, fixtures, migrations, and test harnesses.

Agent-focused bundle for bounded tasks, safe delegation, contract verification, and production evidence.

Safely plans and verifies schema changes, migrations, dirty-data cleanup, and cross-database preconditions.

AI systems bundle for semantic correctness, observability, security, and distributed state.