ue5-automation
This server lets AI agents inspect, build, edit, verify, and save Unreal Engine 5 Blueprints, materials, UMG, and runtime behavior through the editor bridge.
Check editor/bridge health and run diagnostics.
Discover assets and read Blueprint structure, nodes, pins, and connection topology.
Create Blueprints atomically (variables, components, interfaces, events, functions, wiring, compile) with rollback on failure; batch-create multiple Blueprints atomically.
Compile Blueprints and get structured compiler diagnostics with node GUIDs and fix hints.
Search callable nodes/pin signatures, apply whole-graph specs atomically, connect/disconnect pins, and delete nodes with reference checks.
Save or delete assets with scoped saves and read-back verification.
Start/stop PIE, inspect live actors and properties, teleport actors, and capture screenshots for visual self-checks.
Verify runtime behavior and run scripted acceptance scenarios that emit JUnit reports.
Read/filter UE logs incrementally and run whitelisted multi-step batches.
Use materials, UMG, undo, and other advanced capabilities via the generic whitelisted command surface (101 commands).
Track task intent, progress, and retrospectives via task journal tools.
Search tools/commands to discover capabilities and parameters.
Automates Unreal Engine editor workflows, including blueprint creation and editing, material and UMG widget authoring, PIE runtime verification, actor/level management, log diagnostics, and whole-project audits.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@ue5-automation Create a pickup item blueprint with a collision sphere, then verify the pickup logic in PIE."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
UE5 Automation Skill — AI-driven Blueprint Automation for Unreal Engine
English | 简体中文
Let AI agents actually operate the Unreal Editor — node-level blueprint authoring, materials, UMG, and PIE runtime verification, with every write read back and verified.
Works on UE 5.1 – 5.8. Epic's built-in MCP server ships with 5.8+ only — if your project (or your team's mobile line) is on 5.1–5.7, this is the toolkit that covers it. No compilation for 5.1 & 5.8 (prebuilt bridge DLLs included).
Download v6.5.5: portable package · release notes · SkillHub listing
The SkillHub skill package is at 6.5.12 (runtime 6.5.5): doc consistency, single-source tool counts, one calling path, material/UMG/DataTable examples.
Read → write → compile → verify → save/undo. verified=true is the gate.
⭐ If this saves you time, star the repo — it helps other Unreal developers find it.
🎬 Demo GIF — coming soon: a 60–120s capture of "one sentence → agent builds a blueprint → PIE runtime verification" will be embedded here.
Why this exists
Most "AI controls Unreal" attempts die at the same wall: Python cannot reach UEdGraph.
You can demo asset creation in a day, but node-level graph authoring, reliable connections,
and "the operation actually did what it claimed" are where projects stall.
This toolkit solves that with:
A C++ bridge plugin (editor-level UFUNCTIONs): node creation, pin/connection editing with GUID-based targeting, widget-tree authoring, material expression enumeration — the things Python physically cannot do in-engine.
A verification discipline: every write operation is read back and compared (fail-loud, never fake success), connections are verified against live topology, compile results are read back with a log-pattern scan.
Runtime behavior checks: start PIE, read live actor state, teleport with read-back — prove the logic runs, not just that it compiles.
verify_scenariopackages wait/assert/log/screenshot steps into a JUnit-producing acceptance run.Multi-agent safety (v6.5): idempotency keys (retries never double-write), advisory session locks plus task locks, and write receipts with before/after fingerprints.
A security model you can defend to a studio: localhost-only, token auth, a 101-command whitelist (no arbitrary code execution), transaction logs, rollback.
Related MCP server: Unreal-MCP-Ghost
Feature overview
Domain | What it does |
🔍 Blueprint analysis | classes / variables / CDO / node graph / connections → Markdown + Mermaid reports |
🏗️ Blueprint creation | asset + variables + components + interfaces + events + function nodes + wiring + compile (atomic, rolls back on failure) |
✏️ Incremental editing | pin default values, connect/disconnect (topology-verified), delete with reference checks, transaction rollback |
🎨 Materials | create → expressions (params, math, samplers) → connect to properties → compile → instances |
🧩 UMG widget blueprints | create → widget tree (panels/text/buttons…) → set properties → graph logic reuses blueprint commands |
▶️ PIE runtime verification | start/state/live actors/read properties/teleport with read-back/stop |
🖥️ Actors & levels | query / spawn / transform / read properties |
🩺 Log diagnostics | parse engine logs + 19-rule error pattern library + incremental log reader |
📐 Project audit | whole-project scan: counts, complexity, coupling, circular deps, hot spots |
⚡ Batch execution | multi-step operations in a single request |
Security model: localhost bind, token auth, command whitelist, read-back verification on writes, transaction logs, scoped saves only (never saves the whole project).
Relation to Epic's official Unreal MCP (5.8+)
Official MCP (5.8+) | ue5-automation | |
Engine versions | 5.8+ only | 5.1 – 5.8 |
Blueprint graph authoring | object-level tools | node-level (create/wire/verify, GUID-targeted) |
Write verification | returns success | read-back & compare, fail-loud |
Transaction rollback | ✗ | ✓ atomic build, rollback on failure |
Auth | none (loopback only) | token auth + 101-command whitelist + audit log |
Works offline of editor | ✗ | partial (analyzer/log tools) |
On 5.8 | — | integrates with it (ue58-extras adds 8 node-level tools on top) |
On 5.8 the two compose: keep Epic's toolset for scene/asset work, add this bridge for node-level graph authoring and verified writes. On 5.1–5.7, this is the only option.
Requirements
Engine | Read/audit/diagnose | Node-level blueprint writes | Materials/UMG | MCP access |
5.1.x | ✅ out of the box | ✅ prebuilt DLL included | ✅ | ✅ built-in |
5.2–5.5 | ✅ | ⚠️ compile the bridge once ( | ⚠️ same | ✅ |
5.6 / 5.7 | ✅ | ❔ unverified (structurally compatible) | ❔ | ✅ |
5.8 | ✅ | ✅ prebuilt DLL included ( | ✅ | ✅ official MCP + |
Also needs: Python 3.10+ on the host, the Python Editor Script Plugin enabled in the editor, and Remote Execution turned on.
Install (3 steps)
:: 1) dependencies + attempt automatic bridge deployment
install_dependencies.bat
:: 2) manual fallback (project-level, no admin needed):
cd bridge\BlueprintPythonBridge
python deploy.py --target "C:\path\to\YourProject"
:: 3) in the editor: enable "Python Editor Script Plugin",
:: enable Remote Execution (Project Settings → Plugins → Python), restart.
:: Verify:
cd ..\..\skill\ue5-automation\scripts
python ue5_runner.py healthThen one command registers the MCP proxy into your AI clients (auto-detects Claude Desktop / Cursor / Claude Code / VS Code / Windsurf / Codex; every config is backed up before writing):
python skill\ue5-automation\scripts\install_mcp.pyFull guide: README.zh-CN.md (中文) · skill/ue5-automation/DEPLOY.md.
Use it with AI (MCP)
The toolkit is an MCP server. Point any MCP client at it:
UE 5.1–5.5:
http://127.0.0.1:8889/mcp(this repo ships the endpoint; the bridge auto-starts on first write, or seeskill/ue5-automation/docs/MCP-ACCESS.md)UE 5.8: the engine's official MCP server at
http://127.0.0.1:8000/mcp— enable it in project settings (bAutoStartServer=True) and installbridge/ue58-extras/(8 tools; with the prebuilt bridge, node-level graph tools work too)stdio clients (Claude Desktop & co.): launch the bundled stdio proxy — it serves the full tool catalog locally and relays tool calls to the in-editor bridge (token read automatically), so clients that can't set HTTP headers work out of the box:
{ "mcpServers": { "ue5-automation": { "command": "python", "args": ["C:/path/to/ue5-automation/skill/ue5-automation/scripts/ue5_mcp_stdio.py"] } } }
Then just talk to your agent:
"Create a pickup item blueprint with a collision sphere, then verify the pickup logic in PIE."
See MCP-ACCESS.md for client configuration (ZCode / Claude Desktop / Cursor / raw JSON) and the full tool catalog.
Observability: admin dashboard (read-only)
While the agent works, open http://127.0.0.1:8889/ui?token=<contents of the %APPDATA%\ue5_bridge\token file> in a browser to watch what it does — read-only:
a task stream (every tool call with latency and status), a blueprint graph view
(nodes/edges, breathing highlight on recent writes, compile errors pinned to nodes),
a verification wall (verified/warnings cards per write + compile-round counter),
and one-click audit export (zip). Everything is GET-only and can never block the
agent's main path; events are also appended to %APPDATA%\ue5_bridge\sessions\
(JSONL) for after-the-fact audits.
Use it from the command line
cd skill\ue5-automation\scripts
python ue5_runner.py health
python ue5_runner.py read /Game/BP_Test --level L2
python ue5_runner.py build spec.json
python blueprint_editor.py update /Game/BP_Test PrintString InString "Hello"
python blueprint_editor.py connect /Game/BP_Test BeginPlay then PrintString execute
python log_parser.py --log "%PROJECT%\Saved\Logs\YourProject.log" --output diag.json
python analyzer.py --input reader_out.json --summaryThe execution chain (what the agent actually does)
requirements
→ preflight (engine/version/bridge checks)
→ read state (nodes, GUIDs, variables)
→ apply changes (build / edit / wire — every step read-back verified)
→ verify behavior (PIE start → live actor state → teleport w/ read-back → stop)
→ deliver (Markdown + Mermaid reports, transaction logs)
→ clean up (scoped saves only — never saves the whole project)Every write is read back and compared; mismatches raise loud errors. We consider "returns true but did nothing" the cardinal sin of editor automation.
Documentation
Doc | Content |
部署 / 接入 / SOP / 能力 / 排障 | |
中文说明与部署指南 | |
full manual (v0.20) | |
connecting MCP clients to UE 5.1–5.5 | |
19 known pitfalls (and fixes) | |
compiling the bridge for 5.2–5.8 |
Repository layout
├── bridge/ engine-side "hands"
│ ├── BlueprintPythonBridge/ C++ bridge plugin (editor UFUNCTIONs)
│ │ ├── Binaries/Win64/ prebuilt 5.1 DLL
│ │ └── Source/ full source + build_for_engine.bat (5.2–5.8)
│ ├── prebuilt-5.8/ prebuilt 5.8 DLL (+ UnrealEditor.modules)
│ └── ue58-extras/ official-MCP toolset plugin for 5.8 (8 tools)
├── skill/ue5-automation/ host-side brain: bridge client, MCP endpoint,
│ runner, blueprint editor/reader/builder, analyzer,
│ log parser, preflight, docs, report templates
└── CHANGELOG.md per-release summariesQuick success path
Download the portable package from Releases, unzip, run
install.bat, and drag in your.uproject.Restart the editor and confirm
healthreturnsbridge_loaded: true.Build a smoke Blueprint, write
BeginPlay → PrintStringwithapply_graph, compile it withcompile_and_diagnose, and read the pin back withget_pin_default_value.Save the asset and confirm the result. That is the whole loop: write → compile → read back → save.
Full command contract: SKILL.md; SkillHub package: packaging/skillhub/ue5-automation; troubleshooting: TROUBLESHOOTING.md; pitfalls: PITFALLS.md; examples: EXAMPLES.md.
Commercial / Enterprise
The core is Apache-2.0. Paid work is available for enterprise integration, UE 5.2–5.7 bridge adaptation, CI/asset-governance pipelines, training, and custom tooling. Open a GitHub issue titled [Commercial] or contact @miracle1814.
Status & tested-against
Actively maintained — monthly releases, per-release changelog, CI on every push
Offline regression: 852 passed / 0 failed (95 skipped = online-only tests that need a live editor)
Live end-to-end verified on UE 5.1.1 and UE 5.8.2 (editor → official MCP → toolset → C++ bridge → live PIE state)
See CHANGELOG.md for per-release summaries
License
Apache-2.0 — see LICENSE and THIRD_PARTY_NOTICES.md.
Unreal® Engine is a trademark of Epic Games, Inc. This project is an independent developer tool and is not affiliated with or endorsed by Epic Games. Engine headers and binaries remain subject to Epic's EULA; the Apache-2.0 license covers original source code only.
Acknowledgements
Built on top of the UE editor's Python & remote-execution surfaces, and integrating with the official ModelContextProtocol plugin (5.8+).
Available Tools
34 toolsue5_apply_blueprint_graphA
Declarative whole-graph write: one atomic transaction builds/wires the GraphSpec, compiles outside it, verifies by read-back, rolls back on failure. Main path for graph logic; build the spec from ue5_search_blueprint_nodes first. Node-level tools (ue5_connect_pins / ue5_delete_node) only for surgical fixes. | 整图写入主路径:原子事务 + 读回 verified + 失败回滚;先 search 再 apply。
Inputs: bp_path (required), graph_spec (required), replace, session_id, +1 more.
ue5_build_batchA
Create multiple Blueprints in one atomic batch: if any blueprint fails, all assets created by this batch are rolled back. Prefer ue5_build_blueprint unless multi-asset atomicity is actually needed. | 批量创建蓝图(原子化:任一失败回滚本批已建资产)。
Inputs: blueprints (required).
ue5_build_blueprintA
Create a Blueprint from scratch in one atomic call: asset + variables + components + interfaces + events + function nodes + connections + optional compile. If any step fails, assets created by this call are rolled back. spec fields: blueprint_name (or name), package_path, parent_class, variables[{name,type,default_value,category}], components[{name,component_class,properties}], interfaces[str], events[{event_name|name,pos}], functions[{function_name|name, target_class, function_path, defaults:{pin:val}, pos}], connections[{src_node,src_pin,dst_node,dst_pin}], compile_after. Node references accept 32-hex GUIDs (recommended) or titles. | 创建蓝图(原子化:失败回滚已建资产)。
Inputs: spec (required), session_id, idempotency_key.
ue5_check_healthA
Check bridge/editor health: engine version, project name, loaded bridge API list. Call this first to confirm the editor is reachable before any write operation. Returns {engine_version, project, bridge_apis[]}. | 健康检查:引擎版本、项目名、bridge API 清单;任何写操作前先调用。
Inputs: no named parameters.
ue5_compile_and_diagnoseA
Compile a Blueprint AND return structured compiler errors: each error carries asset/function/node_name plus node_guid (the positioning key for incremental edit tools), pin hint when parseable, and a rule-map pitfall_ref/fix_hint. Fields that cannot be parsed are null — never fabricated. Use this instead of ue5_compile_blueprint when a compile fails and you need to locate and fix the offending nodes. fix_hint is advisory only, never a substitute for your own analysis. | 编译蓝图 + 结构化错误诊断(node_guid 定位 + 规则库映射);编译失败需定位修复时用它替代 ue5_compile_blueprint。fix_hint 仅供参考,不代替模型判断。
Inputs: bp_path (required).
ue5_compile_blueprintA
Compile a Blueprint, read back the compile status and scan the log for error patterns. The real compile result is reported: a failed compile returns failure with log evidence, never a fake success. | 编译蓝图 + 回读编译状态 + 日志错误模式扫描。
Inputs: bp_path (required).
ue5_connect_pinsA
Connect pins with triple topology verification (existence, type compatibility, read-back). Node references accept titles or 32-hex GUIDs; ambiguous titles are rejected (fail-loud) — use GUIDs when titles repeat. | 引脚连线(三重拓扑验证;标题歧义 fail-loud 拒绝)。
Inputs: connections (required), blueprint_path (required).
ue5_delete_assetA
Delete an asset with reference checks and read-back verification. Save scope = referencing packages only. | 级联删除资产(删后回读验证;保存作用域=引用者包)。
Inputs: asset_path (required), max_attempts.
ue5_delete_nodeA
Delete a Blueprint node. dry_run=true only checks references without deleting — use it first for nodes that may be referenced elsewhere. | 删除节点(dry_run 仅检查引用,不删除)。
Inputs: node_id (required), blueprint_path (required), dry_run.
ue5_disconnect_pinB
Disconnect every connection attached to one pin of a node, verified against the live topology. | 断开节点某引脚的全部连线(按实时拓扑验证)。
Inputs: node_id (required), pin_name (required), blueprint_path (required).
ue5_execute_commandA
Escape hatch: run any whitelisted bridge command by name — the full 101-command surface, including capabilities without a dedicated tool (materials, UMG widgets, and other advanced operations). Prefer the dedicated tool whenever one exists for the operation; this generic entry point is the last resort and its arguments are less validated. Use ue5_tool_search{query} to find exact command names and their params. 101 commands cover materials/UMG/tasks/undo/PIE/logs/batch. Unknown command → fail-loud拒绝。| 白名单透传;命令名用 ue5_tool_search 查。
Inputs: command (required), params, session_id, idempotency_key.
ue5_list_assetsA
List asset paths under a content folder (default /Game/). Use to discover what exists before reading or editing. recursive=true walks subfolders. | 列出 /Game 资产路径(recursive 控制是否递归)。
Inputs: path (required), recursive.
ue5_list_pie_actorsA
Enumerate live Actors in the PIE world (runtime objects; their paths contain UEDPIE_0_). Filter by actor_class and/or name_filter substring; limit caps the result count. | 枚举 PIE 世界 Actor(运行时对象)。
Inputs: limit, actor_class, name_filter.
ue5_read_blueprintA
Read a Blueprint asset: level L1 = class info + variables + CDO, L2 = node graph + connections, full = both. Read-only and safe to call any time. Returns structured JSON (Markdown/Mermaid reports available via the CLI reader). | 读取蓝图(L1 类信息+变量+CDO / L2 节点+连线 / full)。
Inputs: blueprint_path (required), level, fields, include_cdo, +1 more.
ue5_read_connectionsA
Read the Blueprint's connection topology (which pin connects to which node/pin). Use to verify wiring before and after edits. | 读取蓝图连线拓扑。
Inputs: bp_path (required), fields, graph_name.
ue5_read_logsA
Read the UE log: cursor=0 returns the tail (last lines); cursor>0 returns only new lines after that cursor (incremental polling). Filter by category/pattern. Use for self-service cross-checks of any operation. | 读 UE 日志(cursor=0 取尾部;cursor>0 增量取新增;category/pattern 过滤)。
Inputs: lines, cursor, pattern, category.
ue5_read_nodesA
Read Blueprint nodes: titles, positions, GUIDs and pin defaults. Omit graph_name to search all graphs. Use before connecting pins so you have the GUIDs. | 读取蓝图节点(不传 graph_name = 全图搜索)。
Inputs: bp_path (required), fields, graph_name.
ue5_read_pie_propertyA
Read a runtime Actor property in PIE; omit property_name to list candidate property names first. Use this to prove game logic ran (e.g. health changed after a pickup). | 读 PIE 中 Actor 属性(property_name 空 = 列候选属性名)。
Inputs: actor_name (required), property_name.
ue5_read_pie_stateA
Read PIE state: is_playing, world name, paused, game time. Poll this after ue5_start_pie before reading runtime actors/properties. | PIE 状态:is_playing / 世界名 / 暂停 / 游戏时间。
Inputs: no named parameters.
ue5_run_batchA
Execute a whitelisted-command batch in one request (multi-step task in a single round trip); stops on the first failing step by default. For a single operation prefer the dedicated tool — this is for sequences. | 白名单命令批量执行(单次请求多步;任一步失败默认中止)。
Inputs: steps (required), stop_on_error.
ue5_run_diagnosticsA
Full bridge diagnostics: complete C++ API method matrix and any missing critical APIs. Use when a tool fails with an API-missing error, or after upgrading/replacing the bridge DLL. | Bridge 诊断:全部 C++ API 方法矩阵 + 关键 API 缺失项。
Inputs: no named parameters.
ue5_save_assetA
Save an asset to disk; only_if_is_dirty=true skips clean packages. Saves are scoped to the asset's own package — this tool never triggers a whole-project save. | 保存资产落盘(only_if_is_dirty=true 仅脏包写盘;作用域=该资产包,绝不全工程保存)。
Inputs: asset_path (required), only_if_is_dirty.
ue5_screenshotA
Screenshot for visual self-check (edit -> screenshot -> inspect -> fix). source: editor_viewport | agent_camera (camera {pitch,yaw,roll} required) | pie (fails loudly if PIE not running). Returns an image block + saved_path / camera_used provenance; PNG also saved for the dashboard. | 视觉自检截图;camera_used 可溯源,PIE 未运行 fail-loud。
Inputs: source (required), camera, annotate, out_name, +1 more.
ue5_search_blueprint_nodesA
Search callable nodes in a Blueprint (events / function candidates / flow control / variables) WITH full pin signatures — read-only. Use this BEFORE ue5_apply_blueprint_graph to get accurate targets and pin signatures for the GraphSpec. Engine-library candidates carry names + owning class; their pins are honestly empty (pure-Python channel cannot read UFunction signatures) and say so. | 只读搜索蓝图可调用节点(事件/函数/流程控制/变量候选)+ 完整引脚签名。写 ue5_apply_blueprint_graph 的 GraphSpec 前先调它拿准确签名;引擎库候选如实不带引脚(纯 Python 通道读不到 UFunction 签名)。
Inputs: bp_path (required), limit, query.
ue5_set_pie_transformA
Teleport a PIE Actor at runtime (location and/or rotation, at least one), with read-back verification of the new transform. | 运行时传送 PIE Actor(location/rotation 至少一个;带回读验证)。
Inputs: actor_name (required), location, rotation.
ue5_start_pieA
Start PIE (play/simulate) asynchronously. Poll with ue5_read_pie_state until is_playing=true. Use to verify that logic actually RUNS, not just that it compiles. | 启动 PIE/Simulate(异步;用 ue5_read_pie_state 轮询确认)。
Inputs: mode.
ue5_stop_pieB
Stop the running PIE/Simulate session. | 结束 PIE/Simulate 会话。
Inputs: no named parameters.
ue5_task_beginA
Declare a task for the dashboard: goal, asset, understanding, approach, steps. One active task at a time; close with ue5_task_end. | 声明任务(意图/计划/步骤),面板据此展示脉络。
Inputs: goal (required), asset, steps, approach, +2 more.
ue5_task_endA
Close the active task with a retrospective: status success|failed|aborted, summary (what was implemented), lessons (practical takeaways - what to avoid over-complicating next time, better approaches found). Dashboard shows duration, failed steps, and the retrospective. | 收尾复盘:status 必填;summary 实现了什么;lessons 实操经验(下次避免简单问题复杂化/新方法)。面板自动算总耗时与失败步骤。
Inputs: status (required), lessons, summary, deliverables, +1 more.
ue5_task_listB
List journalized tasks (active + history) with their plans, progress and retrospectives. | 列出任务日志(进行中 + 历史),含计划/进度/复盘。
Inputs: no named parameters.
ue5_task_stepA
Report task progress against the declared plan: step_id (from ue5_task_begin steps) + status running|done|failed|skipped, or a free note when between steps. evidence may carry screenshot paths / error counts for the dashboard. | 上报进度:step_id 定位计划步骤 + status;步骤间可发 note 自由注记;evidence 可带截图路径/错误数作为证据。
Inputs: note, status, step_id, evidence.
ue5_tool_searchA
Search the 101 whitelisted commands and the MCP tools by keyword; returns command names, params and one-line descriptions. Use this before execute_command to discover capabilities without a dedicated tool. | 按关键词检索白名单命令/MCP 工具(参数与一句话说明);调用 execute_command 前先用它查命令名。
Inputs: query (required), limit, include_tools.
ue5_verify_behaviorA
Assert runtime behavior: read a live property and compare with the expected value (exact / tolerance / contains), optionally capture a PIE screenshot as evidence. Requires PIE/Simulate running; mismatch is fail-loud. | 行为验证:读运行时属性与期望比对(支持数值容差/子串),可附截图证据;PIE 未运行会 fail-loud。
Inputs: expected (required), actor_name (required), property_name (required), match, +4 more.
ue5_verify_scenarioA
Run a scripted acceptance scenario in PIE and emit a JUnit report: wait / assert property / screenshot / log_contains / command steps, with automatic PIE start/stop and a total timeout. Destructive commands are rejected unless allow_destructive=true. Returns verified + per-step evidence + junit_path (CI-consumable). | 场景化验收:PIE 内逐步等待/断言/截图/日志/命令,自动起停 PIE,总超时保护;高危命令默认拒绝。verified 是唯一判据。
Inputs: name (required), steps (required), pie_mode, stop_pie, +5 more.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 34 tools
Most tools target a distinct resource+action and the descriptions actively route the agent (apply_blueprint_graph is the 'main path', connect_pins/delete_node are 'surgical fixes'; compile_and_diagnose is the fallback for compile_blueprint). A few boundaries still need the prose to separate: compile_blueprint vs compile_and_diagnose, check_health vs run_diagnostics, verify_behavior vs verify_scenario, and build_blueprint vs build_batch vs apply_blueprint_graph. The 'prefer X unless Y' guidance makes these resolvable, so overlap is minor rather than confusing.
Nearly all tools use the ue5_ verb_noun snake_case pattern (start_pie, read_pie_state, connect_pins, delete_asset, verify_behavior). Minor deviations: noun-first task_list and tool_search break the verb_noun ordering of task_begin/task_step/task_end, and screenshot is a bare noun. The convention is still readable and predictable overall.
34 tools is above the comfortable band and the surface is heavy for an agent to hold in context, with the 101-command escape hatch meaning some dedicated tools are conveniences. However, the domain genuinely spans Blueprint authoring, PIE runtime verification, asset management, acceptance scenarios, tasks, and diagnostics, and each tool maps to a real capability rather than a duplicate. Borderline-heavy but not bloated.
Blueprint coverage is near-complete (search/read/create/apply/wire/disconnect/delete/compile/diagnose), plus full PIE lifecycle and runtime assertion, screenshots, logs, asset list/save/delete, and an execute_command/tool_search escape hatch for materials and UMG. Gaps are minor: no asset rename/duplicate, no pause/resume PIE, and non-Blueprint asset creation only via the generic command path.
Maintenance
Related MCP Connectors
A comprehensive Model Context Protocol (MCP) server that enables AI assistants to control Unreal E…
Control Unreal Engine to browse assets, import content, and manage levels and sequences. Automate…
Make games with AI: one MCP server that connects your agent to Unity, Unreal and Godot editors.
Build and run AI workflows, agents, data tables and 1000+ API integrations from any MCP client.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceEnables AI agents to control the Unity Editor through MCP, allowing scene building, runtime scripting, visual QA, and more.5Apache 2.0
- AlicenseCqualityCmaintenanceEnables AI clients to control Unreal Engine 5 editor for automated Blueprint authoring, level inspection, actor spawning, and other editor workflows via a local Python MCP server and UE plugin.5004AGPL 3.0
- AlicenseNot gradedqualityDmaintenanceEnables AI agents to control and query Unreal Engine 4.27.2 editor through MCP, supporting asset creation, level editing, and project inspection via Python remote execution.129 npm35MIT

Funplay MCP for Unrealofficial
AlicenseNot gradedqualityCmaintenanceEnables AI assistants to directly control and interact with the Unreal Editor through natural language, allowing tasks such as spawning actors, building levels, creating materials, and running Python scripts.11MIT