Synthetic Agent Cost & Reliability Scorecard — mapped agent-run data into reliability, cost-per-success, failure-mode, and break-even views. Demonstration data only; no client result claimed.
https://wrightops-site-pink.vercel.app/#method
1
16
Public Repo Readiness Audit — 50/100 evidence-backed score for Hermes Flight Recorder, with four prioritized fixes. Public demonstration, not client work.
https://github.com/zwright8/hermes-flight-recorder/tree/f917884f00283417004d76498130bd4e6ba82170
1
17
Hermes Flight Recorder — Agent Evaluation & Governance Stack
I built a Python evidence and governance stack that turns tool-using AI-agent runs into normalized traces, scorecards, evidence bundles, dataset exports, and promotion decisions. It supports multiple agent harnesses, deterministic demos, schema validation, redaction, held-out evaluations, and fail-closed quality gates.
https://github.com/zwright8/hermes-flight-recorder