Skip to content

Feedback

UE-MCP includes a built-in feedback system that helps improve tool coverage over time. When an AI agent has to fall back to editor(action="execute_python") because a native tool couldn't handle the task, it can submit structured feedback directly as a GitHub issue.

How It Works

flowchart LR
    Agent[AI Agent] -->|notices tool gap| FT[feedback tool]
    FT -->|GitHub App auth| GH[GitHub Issues]
    GH -->|maintainers triage| Fix[New native tool/action]
  1. During a session, the agent uses editor(action="execute_python") as a workaround for something a native tool should handle
  2. When the task is complete, the agent asks: "I had to use custom Python scripts to get this done. Would you like to submit feedback to improve ue-mcp?"
  3. If the user agrees, the agent calls feedback(action="submit") with details about the gap
  4. A GitHub issue is created automatically on the ue-mcp repository

Privacy

The agent is instructed to strip project-specific details from feedback submissions. Issues should describe the general capability gap, not your project's internals. You can review the issue content before the agent submits it.

Submitting Feedback

The feedback tool has one action:

submit

Parameter Required Description
title Yes Short title describing the tool gap (generic, no project details)
summary Yes What was attempted and why the native tool fell short
pythonWorkaround No The execute_python code used as a workaround
idealTool No What tool/action should handle this natively

Example

feedback(action="submit",
  title="Cannot set default values for Blueprint variables",
  summary="Tried to set a default value on a Blueprint variable. add_variable creates the variable but there's no action to set its default. Had to use execute_python to access the variable's DefaultValue property directly.",
  pythonWorkaround="import unreal; bp = unreal.load_asset('/Game/BP_Player'); ...",
  idealTool="blueprint(action='set_variable_default', assetPath, name, defaultValue)"
)

For Maintainers

Feedback issues are created with the agent-feedback label and include:

  • Summary — what the user was trying to do
  • Ideal Tool/Action — suggested native tool signature
  • Python Workaround — the code that solved it, useful for implementing the native handler

These issues form a prioritized backlog of tool gaps to close.

Claude Code Hooks

If you ran npx ue-mcp init and selected "Agent behavior" hooks, Claude Code will automatically prompt agents to submit feedback whenever they fall back to execute_python. This is a Claude Code PostToolUse hook — it fires deterministically after the tool call, not as a suggestion the agent can ignore.

The hook is configured in your project's .claude/settings.json and calls npx ue-mcp hook post-tool-use under the hood.

Resolving Feedback Issues

Once feedback issues are triaged, anyone can resolve them:

npx ue-mcp resolve <issue-number>

This fetches the issue, creates a branch, launches Claude Code to implement the fix, and opens a PR. See Getting Started for details.