geap-plan-validator

It sends the plan to three remote Vertex AI skeptic models that independently expose dependency defects, hidden assumptions, and missing failure handling before execution begins.

Ask the Skill to review the implementation plan at a specified file path, using your configured Google Cloud project and application default credentials to access the Vertex AI models.

Yes, you need application default credentials, a configured Google Cloud project, and configurable Vertex AI model access to run the remote skeptics and synthesize findings.

First-domino detection identifies the earliest step in an implementation plan whose failure could invalidate the remaining plan, helping teams prioritize corrective changes.

Yes, you can use configurable Gemini or Claude models hosted on Vertex AI as remote skeptics to independently analyze your software engineering milestones, feature work, and refactors.

A synthesis model consolidates the findings from all skeptics, casts an additional validation vote, and confirms only issues supported by at least two votes to ensure reliable plan validation.