SkillAgentSearch skills...

fmodel-mcp

Local MCP server for Unreal Engine asset exploration and export. 32 tools, Codex/ChatGPT desktop installer, powered by CUE4Parse.

Install / Use

claude mcp add plu1337 -- npx -y github:plu1337/fmodel-mcp

If the server publishes to npm under a different name, use that package instead — check the repo README.

About this skill
🔌

MCP Server

Model Context Protocol server

Quality Score

83/100

Supported Platforms

Claude Code
Claude Desktop
OpenAI Codex

Tags

Our assessment of fmodel-mcp

fmodel-mcp scores 83/100 on our quality scale, 645th of 939 AI & Machine Learning skills we index.

Its MCP Server is 8.4 KB long, well organised into 8 sections with 4 code examples: a thorough specification that gives an agent plenty to work with.

It has 10 GitHub stars, so there is little community track record yet; judge it on its content.

Substance
29/30
Structure
20/20
Description
15/15
Adoption
4/20
Freshness
15/15

Maintenance, license and trust

  • The repository was last updated about 2 months ago, so fmodel-mcp is actively maintained.
  • It is released under GPL-3.0, a copyleft license: you can use it, but modified versions you distribute must carry the same license.
  • Its trust signals score 97/100, with no cautions. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.

fmodel-mcp compared with similar skills

All 4 of these similar skills score higher than fmodel-mcp; compare them before choosing.

SkillScoreStarsUpdatedFormat
fmodel-mcp (this skill)by plu1337831059d agoMCP Server
claude-memby thedotmack100112.1ktodayCLAUDE.md
Agent-Reachby Panniantong100108.5k3d agoCLAUDE.md
Understand-Anythingby Egonex-AI10090.3k9d agoCLAUDE.md
headroomby headroomlabs-ai10078.8ktodayCLAUDE.md

Frequently asked questions

How do I install fmodel-mcp?
Run claude mcp add plu1337 -- npx -y github:plu1337/fmodel-mcp. The install tabs above show the steps for each supported agent.
Which AI agents does fmodel-mcp work with?
It is written for Claude Code, Claude Desktop and OpenAI Codex, as a MCP Server file. Other agents that read the same format can often use it too.
Is fmodel-mcp safe to use?
It is GPL-3.0-licensed and scores 97/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
Is fmodel-mcp still maintained?
The repository was last updated about 2 months ago, so fmodel-mcp is actively maintained.
<div align="center">

FModel MCP

Explore Unreal Engine game assets with your AI assistant.

Build Release Windows x64 License: GPL v3

Download · Tool reference · Architecture · Report an issue

</div>

FModel MCP is a local Model Context Protocol server for Unreal Engine archives and assets. It uses the same CUE4Parse parsing and conversion libraries as FModel, exposing 34 tools, four resources/resource templates, and four workflow prompts to MCP clients. No running FModel window, GPU, or Unreal Editor is required.

This is an independent community project, unaffiliated with the FModel or CUE4Parse maintainers.

What you can do

| Task | Available capabilities | | --- | --- | | Open archives | PAK, IoStore, loose assets, engine/game profiles, local mappings, in-memory AES keys | | Reuse FModel setup | Discover saved games and open the last saved selection with local keys, profile overrides and mappings | | Find assets | Directory browsing, paginated path search, asset registry queries, archive statistics | | Understand packages | Exports/imports/names, JSON properties and pointers, references, Blueprint pseudocode | | Examine content | Inline texture previews, localization, registry and binary configuration inspection | | Export assets | Raw files with sidecars, JSON, supported textures, meshes, skeletons, animations, materials, worlds and audio | | Manage work | Background batches, cancellation, progress, per-asset results, manifests and version comparisons |

Actual decoding and conversion depend on the selected game profile, mappings, encryption keys and available codecs. The support matrix and limitations describe the boundaries. The server does not edit or repack game archives.

Install in Codex / ChatGPT desktop

Requires Windows x64. Release ZIPs include .NET 10; you do not need to install the SDK or runtime.

  1. Download fmodel-mcp-1.1.1-win-x64.zip and its .sha256 file from Releases.
  2. Check the ZIP with Get-FileHash .\fmodel-mcp-1.1.1-win-x64.zip -Algorithm SHA256 against the published checksum, then extract it to a folder.
  3. Open PowerShell in that extracted folder and run:
powershell -NoProfile -ExecutionPolicy Bypass -File .\Install-Mcp.ps1

The installer copies the complete release to %LOCALAPPDATA%\FModelMcp, tests its MCP connection, and registers fmodel using the Codex CLI (including the CLI bundled with the desktop app). On a fresh install it detects FModel's saved selection, allows that game folder and its local mapping cache, and connects an existing cached Oodle library when available. Saved AES keys are read directly from FModel's settings when you open a saved game; they are never copied into the MCP configuration. An Inputs folder is also available for additional archives; exports go to a separate Exports folder. To allow an existing game folder instead, run the installer from PowerShell with:

.\Install-Mcp.ps1 -InputRoots 'D:\Games\MyGame\Content\Paks', 'D:\Mappings'
  1. Restart fmodel in Settings → MCP servers, or restart the desktop app. Type /mcp to check it.
  2. Ask your assistant:

Use FModel to show its capabilities and the allowed input folders. Help me select the right engine profile, open my archives, and find texture assets.

If you already use FModel, simply ask:

Open my saved FModel game and show me what's available.

The assistant can call fmodel_list_saved_games, then fmodel_open_saved_game with no arguments to open the last saved selection. This reads saved settings; it does not inspect unsaved desktop state. When multiple cached mappings match, the assistant must select the correct game build explicitly.

The desktop app, CLI, and IDE extension share local MCP configuration. ChatGPT web does not read this local configuration; this release is a local stdio server.

Configuration, updates and removal

Edit %LOCALAPPDATA%\FModelMcp\config.json to change allowed inputRoots, outputRoot, limits, or the path to a locally supplied Oodle library. Restart the server after changes. Input directories must exist. Mapping files must be inside an allowed input root.

Run the installer from a newer extracted release to update. Existing settings are preserved unless you pass replacement folder parameters, in which case the previous configuration is backed up. Versioned installation directories keep a running server from interrupting file copies.

To connect an existing MCP installation to FModel, or refresh allowed folders after changing games, run ./Install-Mcp.ps1 -ImportFModel from an extracted release. Use -FModelSettingsPath for an alternate settings file. Changes to saved keys/profile overrides are read on each saved-game open; changed access roots or codec paths require a server restart. Existing sessions keep their original profile.

To remove the registration, run codex mcp remove fmodel, or remove it in the app's MCP settings. Then remove unwanted installation folders yourself; keep any inputs or exports you need.

Other MCP clients

Run Install-Mcp.ps1 -SkipRegistration and use the generated %LOCALAPPDATA%\FModelMcp\mcp-client.json, or configure the executable directly:

{
  "mcpServers": {
    "fmodel": {
      "command": "C:/Tools/fmodel-mcp/FModel.Mcp.exe",
      "args": ["--config", "C:/Tools/fmodel-mcp/config.json"]
    }
  }
}

Keep all extracted files together. Use absolute paths and copy/edit the example server configuration. Starting the executable manually waits for MCP input; that is expected.

Example workflows

| Ask your assistant | Tool sequence | | --- | --- | | “Find and preview the game's logo.” | capabilities → list_options → open_game → search_assets → inspect_package → preview_texture | | “Export these meshes as glTF.” | inspect_package → start_export → job_status → job_results | | “Explain this Blueprint's data and script.” | open_game with readScriptData → inspect_package → get_properties → decompile_blueprint | | “What changed between these two builds?” | open both builds → compare_sessions → inspect changed packages |

All tool names begin with fmodel_. The server supplies model instructions and prompts that teach the sequence, pagination, profile selection, session cleanup and export-result checks. See the full AI command plan and argument examples.

Build from source

Install the .NET 10 SDK and PowerShell 7, then:

git clone https://github.com/plu1337/fmodel-mcp.git
cd fmodel-mcp
./scripts/Build-Mcp.ps1

The script downloads a checksum-verified, pinned CUE4Parse source revision, applies the documented host integrations, restores NuGet packages, runs the tests, and creates a self-contained ZIP with notices and its SHA-256 checksum under artifacts/. The downloaded dependency and build products are ignored by Git. Optional upstream C++ codec compilation is skipped; see the codec setup guide.

Tests cover real PAK and IoStore fixtures, encrypted-index mounting, package inspection, texture preview, localization, exports, cancellation, path confinement, and the official MCP client protocol. See CONTRIBUTING.md for focused development commands.

License and credits

The server is distributed under GPL-3.0. FModel inspired the workflow; CUE4Parse provides the parsing and conversion engine under Apache-2.0. Dependencies retain their own licenses. Release archives include generated third-party notices and bundled license files. Releases also include a source bundle with the pinned CUE4Parse source and the applied integrations. No game archives, mappings, AES keys, or proprietary Oodle DLLs are included in the runtime release.

Related Skills

View on GitHub
GitHub Stars10
CategoryAI
Updated1mo ago
Forks0

Languages

C#

Trust signals

97/100

From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.

1 info