Loading ToolBench
Give coding agents eyes and hands in Unreal Engine with a plugin, MCP bridge, and skills built to cut token-heavy guesswork
Server has 21 well-named tools with consistent verb_noun patterns (ue_get_*, ue_set_*, ue_spawn_*, ue_destroy_*). All tools have descriptions (avg ~80 chars, within baseline 34-392). Input schemas are present and typed for all tools. However, parameter descriptions are sparse or missing in several tools (e.g., ue_get_property target param lacks clarity on which field to use), and output schemas are not documented. Error handling exists but lacks recovery guidance. No tool annotations (readOnlyHint/destructiveHint) despite clear risk classifications in metadata.
Search for assets in the Unreal Engine content browser
Capture a screenshot of a specific actor
Destroy an actor in the level with safety checks
Frame an actor in the viewport by moving the camera to focus on it
Get the current debug draw state including lines and points
Get editor diagnostics including compilation warnings and errors
Get the current state of the Unreal Editor including project name, current map, and PIE status
Output schemas not documented. Tools return results but LLMs cannot predict field structure, forcing trial-and-error parsing. E.g., ue_get_level_actors returns a list but fields are not specified.
Target parameter (objectPath vs actorName) lacks clarity. ue_get_property, ue_set_property, ue_select_actor_safe, ue_destroy_actor_safe, ue_frame_actor all accept a 'target' object with two optional fields but do not explain which is required or how they interact.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite clear risk classifications in metadata. Tools marked DESTRUCTIVE (ue_destroy_actor_safe) or WRITE should declare annotations to guide agent planning.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 59 | <=2025-11-25 | v2 |
Query actors in the current level with optional filtering by class name or name substring
Get the status of Live Coding in the Unreal Editor
Retrieve output log entries from the Unreal Editor
Read a property value from an actor or object in Unreal Engine
Get a list of currently selected actors in the Unreal Editor viewport
Get the current viewport camera position and rotation
Capture a screenshot of the current viewport
Check the health and readiness of the UE Agent Bridge backend connection
Run a console command with safety checks against an allowlist
Select an actor in the editor with safety checks
Write a property value to an actor or object in Unreal Engine
Set the viewport camera position and rotation
Spawn an actor in the level at the specified location and rotation with safety checks
Trigger a Live Coding build with safety checks
Error responses lack recovery guidance. errorResponse() returns plain text but does not categorize errors as retryable, user-fixable, or fatal, nor suggest next steps (e.g., 'Try search_users() first').
ue_run_console_command_safe accepts only commandId from allowlist but does not document what commands are available or how to discover them. LLMs cannot infer valid commandIds.