Multi Agent Workflow Manager

Overview

This agent acts as a governance layer for complex software development tasks. It enforces a structured, multi-stage workflow to ensure that changes are small, testable, and properly reviewed before implementation.

It dictates the sequence of applying specific 'skills' (like team-contract or system-architect) and invoking specialized 'subagents' (like backend-engineer or observability-engineer). The core principle is to prevent large, monolithic changes by enforcing handoffs and architectural sign-offs.

Capabilities

• Workflow Enforcement: Guides users through the correct sequence of actions for any given change.
• Architectural Review: Mandates running the system-architect-gatekeeper for non-trivial or cross-boundary modifications.
• Boundary Control: Utilizes an orchestrator to classify scope and explicitly manage handoffs between different engineering domains (e.g., frontend vs. backend).
• Quality Gates: Ensures that every change incorporates a manual test checklist, rollback notes (team-contract), and necessary telemetry setup (observability-engineer).

Example Use Cases

1. Implementing a New Feature: Start by running system-architect for an RFC. Then, use the orchestrator to break down the work into small, sequential tasks (e.g., Backend API first, then Frontend integration).
2. Updating Deployment Process: Apply the vpn-suite skill for layout changes, followed by invoking the ci-watcher subagent to update CI/CD pipelines.
3. Monitoring a Service Change: After any core implementation, always invoke the observability-engineer to ensure metrics, logging, and SLOs are correctly implemented before merging.