HammerTime-MCP is an MCP integration for HammerTime. Enabling AI-assistance in mapping.

HammerTimeMCP
HammerTimeMCP is a Model Context Protocol (MCP) plugin for HammerTime. It lets AI assistants connect to the running HammerTime editor, inspect maps, and perform editor actions through a controlled tool interface.
The project keeps HammerTime as the source of truth. The AI client talks to hammertime-mcp.exe, and the CLI forwards requests to the HammerTime plugin over a local authenticated named pipe.
AI client -> hammertime-mcp.exe -> named pipe -> HammerTime.Mcp.Plugin -> HammerTime editor
Features
- Inspect open documents, selections, entities, brushes, textures, faces, and map bounds.
- Create and edit brushes, entities, brush entities, prefabs, overlays, cordons, and texture data.
- Move, rotate, scale, delete, select, focus, highlight, and search map objects.
- Run map validation, problem checks, safe fixes, leak pointfile loading, compile profiles, undo, and redo.
- Configure common MCP clients through the included installer.
Expectations:
HammerTime-MCP can create playable maps, but it should not be treated as a complete scratch to finish map-making solution. AI is not yet advanced enough to reliably create highly detailed worlds on its own. However, with clear directions and your own manual edits, it can help you create maps much faster. Prompt quality is very important. The better your instructions are, the better the result on the maps, brushes, and layouts will be.
In our current testing, Gemini 3.5 Flash has performed best at creating brushes. Claude Opus 4.7/4.8 comes second, while ChatGPT has been mixed, ranging from acceptable to poor for brush creation.
However, when it comes to fixing scripting issues, GPT-5.5 has performed very well.
Is this the fault of the MCP? No. As mentioned earlier, current AI models are not yet good enough to create highly detailed worlds by themselves. This will likely improve with future AI models, which should be able to create more visually appealing and complex levels.
For now, HammerTime-MCP works best as a map-making assistant: it can speed up your workflow, generate useful starting points, and help with scripting or layout ideas, but the best results still come from combining AI generation with human direction and editing.
Watch HammerTimeMCP in action:
Project Layout
HammerTime.Mcp.Cli\ MCP stdio server and installer commands
HammerTime.Mcp.Plugin\ HammerTime editor plugin and named-pipe bridge
HammerTime.Mcp.Shared\ Shared protocol, DTOs, config, and tool catalog
HammerTime.Mcp.Tests\ MSTest coverage for bridge/tool metadata
MCP-Install\ Generated install bundle
Requirements
- Windows.
- HammerTime installed or built locally.
- .NET 8 Runtime for the MCP server.
- Visual Studio 2022 or the .NET SDK to build from source.
Build
dotnet build .\HammerTime.MCP.sln -c Debug /p:Platform="Any CPU"
This refreshes the install bundle under MCP-Install\Server and MCP-Install\Plugin.
If HammerTime is in a custom location, pass its editor output path:
dotnet build .\HammerTime.MCP.sln -c Debug /p:Platform="Any CPU" /p:HammerTimeEditorOutput="C:\Path\To\HammertimeEditor\"
Installation & Usage
Download the zip file in Releases
Run the bundled installer:
.\MCP-Install\install.bat
Make sure HammerTime is not running while installing.
The installer can configure several MCP clients, including Codex, Cursor, VS Code, Claude, Windsurf, OpenCode, Kimi Code, Antigravity, and Gemini CLI.
Installer should auto-detect HammerTime installation path but if path is not detected automatically:
.\MCP-Install\install.bat codex user "C:\Program Files (x86)\HammertimeEditor"
After a successful install, Run HammerTime and restart your AI CLI or IDE.
Verify
With HammerTime running, check the bridge:
.\MCP-Install\Server\hammertime-mcp.exe status
For diagnostics:
.\MCP-Install\Server\hammertime-mcp.exe doctor
Run tests:
dotnet test .\HammerTime.MCP.sln -c Debug /p:Platform="Any CPU"
Notes
The bridge config is stored at:
%APPDATA%\HammerTime.MCP\config.json
HammerTimeMCP controls the live editor state. Review AI-driven changes before saving important maps, and keep backups or version control for serious work.