Drive Unreal Engine 5 from an AI agent: Remote Control API + in-editor Python, no C++ plugin to compile
Three tools with Italian descriptions and minimal English documentation. No output schemas documented for any tool. No error handling guidance. No tool annotations (readOnlyHint/destructiveHint). Risk classification present (READ_ONLY, WRITE) but not exposed to LLM.
Elenca le installazioni di Unreal Engine trovate sulla macchina. Cerca in: variabile UE_MCP_ENGINE_DIRS, elenco dell'Epic Games Launcher, registro di Windows. Da chiamare per prima quando non c'è ancora un progetto.
Elenca i template ufficiali disponibili nell'installazione (TP_Blank, TP_ThirdPerson, ...).
Creates a new Unreal Engine project from specification
Tool descriptions in Italian; LLMs trained on English will not understand intent. ue_engine_list, ue_engine_templates, ue_project_create all have Italian descriptions that block LLM comprehension.
ue_engine_list and ue_engine_templates have empty input schemas ({}).
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields without knowing response structure.
ue_project_create parameters have minimal descriptions (1-3 words: 'Project name', 'Directory where project will be created'). No format constraints, ranges, or dependencies documented.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 42 | 2026-07-28+ | v2 |
No error handling guidance. Tools do not indicate what errors are retryable, user-fixable, or fatal. No recovery suggestions for common failures (engine not found, project creation failed).