KotorMCP
MCP server for interacting with Knights of the Old Republic game installations, enabling resource discovery, inspection, and analysis for AI agents.
README
<div align="center">
KotorMCP
Model Context Protocol server for Knights of the Old Republic game resources
</div>
A Model Context Protocol (MCP) server that exposes context-rich tools for AI agents to interact with Knights of the Old Republic (KOTOR) and Knights of the Old Republic II: The Sith Lords (TSL) game installations. This server provides intelligent resource discovery, installation management, and game data inspection capabilities tailored for automated analysis and debugging workflows.
Features
-
Installation Detection & Management
- Automatic detection of KOTOR 1 and KOTOR 2 installations
- Environment variable support (
K1_PATH,K2_PATH,TSL_PATH, etc.) - Default path discovery (Windows registry, common install locations)
- Installation caching for performance
-
Resource Discovery & Inspection
- List resources from all locations (override, modules, chitin, streams)
- Filter by resource type, name patterns, and module scope
- Deep resource summarization (GFF structures, 2DA tables, TLK strings)
- Resource metadata extraction (size, location, type)
-
Journal & Plot Analysis
- Comprehensive journal entry overview (
global.jrl) - Plot category organization
- Quest entry enumeration
- Cross-references with game scripts and dialogs
- Comprehensive journal entry overview (
-
AI-Optimized Workflows
- Context-rich responses designed for LLM consumption
- Structured JSON output for programmatic access
- Efficient resource scanning with configurable limits
- Installation-aware resource resolution
Installation Guide
Prerequisites
- Python 3.8+
- A valid KOTOR 1 or KOTOR 2 installation
- MCP-compatible client (Claude Desktop, Cursor, etc.)
Quick Start
Using uv (Recommended)
# End users: run with --refresh for latest (no install needed)
uvx --refresh kotormcp
# Developers: run from local source with --with-editable
uvx --with-editable Libraries/PyKotor --with-editable Tools/KotorMCP kotormcp
uv run --directory Tools/KotorMCP/src --module kotormcp
# Or install editable
uv pip install -e Tools/KotorMCP
Using pip
pip install kotormcp
# Or from source: pip install -e Tools/KotorMCP
Configuration
Claude Desktop
Add the following to your Claude Desktop configuration file:
Windows: %APPDATA%\Claude\claude_desktop_config.json
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Linux: ~/.config/Claude/claude_desktop_config.json
{
"mcpServers": {
"kotormcp": {
"command": "uvx",
"args": [
"--refresh",
"--from",
"kotormcp @ git+https://github.com/th3w1zard1/KotorMCP.git",
"kotormcp"
],
"env": {
"K1_PATH": "C:\\Program Files\\steamapps\\common\\swkotor",
"K2_PATH": "C:\\Program Files\\steamapps\\common\\Knights of the Old Republic II"
}
}
}
}
Cursor / VS Code (Claude Dev Extension)
Add to your MCP configuration:
{
"mcpServers": {
"kotormcp": {
"command": "uvx",
"args": [
"--refresh",
"--from",
"kotormcp @ git+https://github.com/th3w1zard1/KotorMCP.git",
"kotormcp"
],
"env": {
"K1_PATH": "C:\\Program Files\\steamapps\\common\\swkotor",
"K2_PATH": "C:\\Program Files\\steamapps\\common\\Knights of the Old Republic II"
}
}
}
}
Environment Variables
The server automatically detects installations using these environment variables (in order of precedence):
KOTOR 1:
K1_PATHKOTOR_PATHKOTOR1_PATH
KOTOR 2:
K2_PATHTSL_PATHKOTOR2_PATHK1_PATH(fallback)
If environment variables are not set, the server will attempt to find installations using default paths (Windows registry, common install locations).
Usage Guide
The server provides a growing set of installation and discovery tools for interacting with KOTOR installations:
1. detectInstallations
Detect available KOTOR installations and their paths.
Parameters: None
Response:
{
"K1": [
{
"path": "C:\\Program Files\\LucasArts\\SWKotOR",
"exists": true,
"label": "default"
}
],
"K2": [
{
"path": "C:\\Program Files\\LucasArts\\SWKotOR2",
"exists": true,
"label": "env"
}
]
}
Example:
use_mcp_tool({
server_name: "kotormcp",
tool_name: "detectInstallations",
arguments: {}
})
2. loadInstallation
Load and cache a KOTOR installation for subsequent operations.
Parameters:
game(string, required): Game identifier ("k1","k2","tsl","kotori","kotor2")path(string, optional): Explicit installation path (overrides environment variables)
Response:
{
"game": "K1",
"path": "C:\\Program Files\\LucasArts\\SWKotOR"
}
Example:
use_mcp_tool({
server_name: "kotormcp",
tool_name: "loadInstallation",
arguments: {
game: "k1",
path: "C:\\Program Files\\LucasArts\\SWKotOR"
}
})
openInstallation
Build or reuse a compacted in-memory installation snapshot and return a snapshot handle for follow-up queries.
Parameters:
game(string, required): Game identifier ("k1","k2","tsl","kotori","kotor2")path(string, optional): Explicit installation path (overrides environment variables)refresh(boolean, optional): Force the snapshot to rebuild instead of reusing the cached snapshot for the same game/path
Response:
{
"snapshotId": "2d6bf3f5108b4f498cb0d7147d6e31b9",
"cached": false,
"game": "K1",
"path": "C:\\Program Files\\LucasArts\\SWKotOR",
"policy": "default",
"resourceCount": 84217,
"omittedPayloadCount": 19042
}
getInstallationSnapshot
Page through a snapshot handle returned by openInstallation and optionally include compacted per-resource documents.
Parameters:
snapshotId(string, required): Snapshot handle returned byopenInstallationresourceTypes(array of strings, optional): Filter by resource type (for exampleNSS,DLG,TLK)resrefQuery(string, optional): Filter by resref or filename substringsourceQuery(string, optional): Filter by source or container path substringincludeData(boolean, optional): Return compacted per-resource documents instead of metadata-only summarieslimit(number, optional): Maximum number of results (default: 50)offset(number, optional): Skip the first N filtered results (default: 0)
Response:
{
"snapshotId": "2d6bf3f5108b4f498cb0d7147d6e31b9",
"total": 1,
"offset": 0,
"limit": 50,
"nextOffset": null,
"includeData": true,
"items": [
{
"resource": "hello.nss",
"restype": "NSS",
"encoding": "text",
"data": "void main() {}\n"
}
]
}
getInstallationGraph
Page through canonical dependency edges extracted from a snapshot handle returned by openInstallation.
Parameters:
snapshotId(string, required): Snapshot handle returned byopenInstallationedgeKinds(array of strings, optional): Filter by edge kind such asscript,conversation, ortemplate_resreftargetTypes(array of strings, optional): Filter by target resource type such asNSSorDLGquery(string, optional): Filter by target name, source resource, or field path substringsourceQuery(string, optional): Filter by source document path or resource path substringlimit(number, optional): Maximum number of results (default: 50)offset(number, optional): Skip the first N filtered results (default: 0)
Response:
{
"snapshotId": "2d6bf3f5108b4f498cb0d7147d6e31b9",
"total": 3,
"offset": 0,
"limit": 50,
"nextOffset": null,
"items": [
{
"sourceDocumentPath": "Override/fixture.utp.json",
"edgeKind": "script",
"targetName": "open_script",
"targetRestypes": ["NCS", "NSS"],
"targetResolved": true,
"targetDocumentPaths": ["Override/open_script.nss.json"],
"fieldPath": "OnOpen"
}
]
}
3. listResources
List resources from the active installation with filtering options.
Parameters:
game(string, required): Game identifierlocation(string, optional): Resource location filter ("all","override","modules","chitin","streams") - default:"all"moduleFilter(string, optional): Filter by module name (e.g.,"001ebo")resourceTypes(string, optional): Comma-separated resource type extensions (e.g.,"gff,dlg,jrl")resrefQuery(string, optional): Filter resources by name pattern (case-insensitive)limit(number, optional): Maximum number of results (default: 50)
Response:
{
"count": 25,
"items": [
{
"resref": "global",
"restype": "JRL",
"source": "override",
"size": 12345,
"module": null
}
],
"truncated": false
}
Example:
use_mcp_tool({
server_name: "kotormcp",
tool_name: "listResources",
arguments: {
game: "k1",
location: "override",
resourceTypes: "jrl,gff",
resrefQuery: "global",
limit: 10
}
})
4. describeResource
Get detailed information about a specific resource, including structured data for GFF files, 2DA tables, and TLK files.
Parameters:
game(string, required): Game identifierresref(string, required): Resource name (without extension)restype(string, required): Resource type extension (e.g.,"jrl","gff","2da","tlk")order(array, optional): Search location priority order (default:["override", "custom_folders", "modules", "chitin"])
Response:
{
"resref": "global",
"restype": "JRL",
"source": "override",
"size": 12345,
"summary": {
"type": "JRL",
"categories": 5,
"entries": 42,
"structure": "..."
}
}
Example:
use_mcp_tool({
server_name: "kotormcp",
tool_name: "describeResource",
arguments: {
game: "k1",
resref: "global",
restype: "jrl"
}
})
5. journalOverview
Get a comprehensive overview of journal entries and plot categories from global.jrl.
Parameters:
game(string, required): Game identifier
Response:
{
"count": 5,
"categories": [
{
"id": 0,
"name": "Main Quest",
"entries": [
{
"id": 0,
"title": "Escape from Taris",
"text": "You must escape from the planet Taris..."
}
]
}
]
}
Example:
use_mcp_tool({
server_name: "kotormcp",
tool_name: "journalOverview",
arguments: {
game: "k1"
}
})
Common Workflows
Finding All Resources That Modify Plot Points
// 1. Load installation
await loadInstallation({ game: "k1" });
// 2. List all journal-related resources
const resources = await listResources({
game: "k1",
resourceTypes: "jrl,gff,dlg",
resrefQuery: "global",
limit: 100
});
// 3. Get journal overview
const journal = await journalOverview({ game: "k1" });
// 4. Search for scripts that reference plot points
const scripts = await listResources({
game: "k1",
resourceTypes: "ncs",
resrefQuery: "plot",
limit: 50
});
Investigating Module Structure
// 1. List all resources in a specific module
const moduleResources = await listResources({
game: "k1",
location: "modules",
moduleFilter: "001ebo",
limit: 200
});
// 2. Get detailed information about key resources
const area = await describeResource({
game: "k1",
resref: "001ebo",
restype: "are"
});
const git = await describeResource({
game: "k1",
resref: "001ebo",
restype: "git"
});
Debugging Missing Resources
// 1. Check all locations for a resource
const resource = await describeResource({
game: "k1",
resref: "missing_resource",
restype: "gff",
order: ["override", "modules", "chitin", "streams"]
});
// 2. List similar resources
const similar = await listResources({
game: "k1",
resrefQuery: "missing",
limit: 20
});
Architecture
KotorMCP is built on the Model Context Protocol specification and uses the official Python MCP SDK. The server:
- Caches installations for performance across multiple tool calls
- Resolves resources using the same logic as
PyKotorCLIand PyKotor'sInstallationclass - Provides structured summaries optimized for LLM consumption
- Follows established patterns from engine reimplementations (reone, xoreos, kotor.js)
Implementation Notes
- Resource scanning logic mirrors
scripts/investigate_module_structure.py - Journal summarization follows the structure documented in
vendor/xoreos/src/engines/nwn2/journal.cpp - Resource resolution uses PyKotor's
Installationclass with configurable search order - GFF structure summarization provides hierarchical field overviews
Development
Building from Source
# Clone the repository
git clone https://github.com/OpenKotOR/PyKotor.git
cd PyKotor
# Install dependencies
uv pip install -e Tools/KotorMCP
# Run the server
python -m kotormcp.server
Project Structure
Tools/KotorMCP/
├── src/
│ └── kotormcp/
│ ├── __init__.py
│ └── server.py # Main MCP server implementation
├── pyproject.toml # Project metadata and dependencies
├── requirements.txt # Pip-compatible requirements
└── README.md # This file
Dependencies
mcp>=0.1.1- Model Context Protocol Python SDKpykotor>=2.3.2- PyKotor core library for KOTOR file format support
Contributing
Contributions are welcome! Please:
- Fork the repository
- Create a feature branch
- Make your changes
- Submit a pull request
For bug reports or feature requests, please open an issue on GitHub.
License
This project is part of the PyKotor ecosystem and is licensed under the LGPL-3.0-or-later License. See the main repository LICENSE file for details.
Related Projects
- PyKotor - Core library for KOTOR file formats
- PyKotorCLI - Command-line interface for KOTOR resources
- Holocron Toolset - GUI editor for KOTOR files
Acknowledgments
- Built on the Model Context Protocol specification
- Uses the PyKotor library for game file format support
- Inspired by engine reimplementations: reone, xoreos, kotor.js
<div align="center">
Made with ❤️ for the KOTOR modding community
Report Bug · Request Feature · Documentation
</div>
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.