Automate end-to-end benchmarking of Unreal Engine AI agents with unified scoring.
npx skills add https://github.com/blackplume233/UnrealMCPHub --skill ue-benchmark
Please help me install this Agent Skill. Skill: ue-benchmark Source: https://github.com/blackplume233/UnrealMCPHub/tree/main/skills/ue-benchmark Command: npx skills add https://github.com/blackplume233/UnrealMCPHub --skill ue-benchmark
UE Agent Benchmark provides a unified end-to-end evaluation framework that measures AI agents driving Unreal Engine prototypes from initialization to playable outcomes, delivering objective scoring across multiple scenarios.
Run the ue-benchmark framework against a configured Unreal project and a scenario (e.g., vampire-survivors-v1) to generate initial scores and validation artifacts.
High-intent search queries and answers about installing and using this skill.
You can benchmark Unreal Engine AI agents by running automated end-to-end playthroughs against configured Unreal projects using scenario files. This framework provides a unified scoring system that evaluates playability, packaging, and AI reviews to generate objective results.
Scoring metrics for evaluating Unreal Engine AI playability include PackageGate, PVP, ContentScore, and AIReviewScore. These metrics measure initialization through playable outcomes, while also tracking token-based costs for comprehensive agent evaluation.
Yes, you can run automated AI agent evaluations in CI-like environments for Unreal projects. The framework provides automation-ready prompts, evaluation protocols, and artifact validation to enable reproducible benchmarking across multiple scenes and configurations.
Multi-scenario benchmarking is handled through a library of scenario files, such as vampire-survivors-v1, located under the scenarios directory. This setup supports diverse game styles and allows future expansions by applying unified scoring across different configurations.
You do not need to manually package Unreal projects before running an AI agent benchmark. The framework automates end-to-end evaluation from initialization to playable outcomes, applying PackageGate checks to verify packaging as part of the scoring process.
sipherxyzAutomate Unreal Engine 5 C++ builds with Live Coding or Unreal Build Tool.
OpenCovenAutomate AI prompt selection and optimization to reduce token costs.
Arsia-MonsAutomate Silencer menus, state capture, and gameplay validation via CLI.
LeKayScaffold test directory structures and GitHub Actions CI/CD pipelines for Godot, Unity, and Unreal Engine projects.
