interview-simulation-reviewer

To evaluate an AI-driven interview simulation, you can analyze transcript quality, signal integrity, strategy alignment, and knowledge-graph health. This process provides structured assessment across turns to deliver actionable insights from the simulation outputs.

Yes, you can supply an optional scoring CSV alongside your JSON simulation outputs to enrich the interview review. The reviewer incorporates this CSV data into its five-part analysis to perform scoring decomposition and validate results against methodology directives.

Knowledge-graph health is one of the five analytical dimensions evaluated during interview signal diagnostics. It assesses the consistency and integrity of the graph structure across turns to ensure the AI-driven simulation maintains coherent strategy alignment throughout the transcript.

The interview simulation analysis report is saved as a consolidated Markdown file. This structured report is automatically exported to the synthetic_interviews directory, specifically named review_<json_filename_without_extension>.md, ensuring full traceability across cross-run validations.

You perform consistent cross-run validation by feeding JSON simulation outputs to the reviewer alongside methodology directives. It generates structured Markdown reports saved to the synthetic_interviews directory, allowing you to compare transcript quality, signal integrity, and knowledge-graph health across multiple runs.