What problem does it solve?
Applying multiple review or analysis skills to one task in a single context produces shallow, overlapping results with no independent verification. This Skill turns a library of installed skills into a coordinated multi-agent graph where each sub-agent node loads exactly one skill, works a bounded scope, and returns schema-shaped findings that are deduplicated, adversarially verified, and synthesized into one deliverable.
Core Features & Use Cases
- Skills-as-nodes orchestration: Map responsibilities to installed skills one per node, fan them out in parallel, and keep each lens isolated in its own context.
- Structured contracts and verification: Every stage returns schema-shaped findings with file:line evidence and fixed severities; independent verifier nodes attempt to refute each finding before it is reported.
- Three runtimes: Prefer the Workflow tool (dynamic workflows with enforced schemas and resume), fall back to Agent-tool fan-out, or run a labeled sequential degraded mode.
- Use Case: Reviewing a large pull request through five architectural lenses at once — scout the diff inline, fan out one node per skill, dedup findings, verify each adversarially, and deliver a severity-ranked report stating what was covered and what was not.
Quick Start
Ask the agent to use graph engineering to fan out one sub-agent per relevant skill over this change set, verify the findings adversarially, and synthesize a single ranked report.
