AGS Extend SDK MCP Server
oficialAn MCP server to help AI assistants to answer questions and generate AccelByte Extend SDK code more effectively .
AGS Extend SDK MCP Server
This Model Context Protocol (MCP) server exposes Extend SDK functions and models as additional context to language models. It helps AI coding assistants and other MCP clients to answer questions and generate Extend SDK code by providing the following tools.
search-symbols– Search for symbols (functions and models) by name, tags, description (fuzzy)describe-symbols– Get detailed information about specific symbols by their IDscreate-extend-app– Prompt template for creating Extend app projects
Quickstart
Prerequisites
- Cursor
- Docker
[!NOTE] The instructions below can be adapted for other MCP clients as well e.g. Claude Desktop, Gemini CLI, and Visual Studio Code.
Alternative 1: Using STDIO transport (default)
-
Pull the AGS Extend SDK MCP Server container image. For example, with image tag 2026.2.0.
docker pull ghcr.io/accelbyte/ags-extend-sdk-mcp-server:2026.2.0 -
Switch to your project directory and create
.cursor/mcp.jsonwith the following content.{ "mcpServers": { "extend-sdk-mcp-server": { "command": "docker", "args": [ "run", "-i", "--rm", "-e", "CONFIG_DIR", "ghcr.io/accelbyte/ags-extend-sdk-mcp-server:2026.2.0" ], "env": { "CONFIG_DIR": "config/go" } } } }The
CONFIG_DIRvalue above is for Go Extend SDK. For other Extend SDK languages, see here. -
Open your project directory in Cursor and open
File>Preferences>Cursor Settings, InCursor Settings, clickMCP, and make sureextend-sdk-mcp-serveris enabled.
Alternative 2: Using Streamable HTTP transport
-
Pull the AGS Extend SDK MCP Server container image. For example, with image tag 2026.2.0.
docker pull ghcr.io/accelbyte/ags-extend-sdk-mcp-server:2026.2.0 -
Start the MCP server with streamable HTTP transport.
docker run -p 3000:3000 \ -e TRANSPORT=http \ -e PORT=3000 \ -e CONFIG_DIR=config/go \ -e NODE_ENV=production \ -e LOG_LEVEL=info \ ghcr.io/accelbyte/ags-extend-sdk-mcp-server:2026.2.0The
CONFIG_DIRvalue above is for Go Extend SDK. For other Extend SDK languages, see here. -
Switch to your project directory and create
.cursor/mcp.jsonwith the following content.{ "mcpServers": { "extend-sdk-mcp-server": { "url": "http://localhost:3000/" } } } -
Open your project directory in Cursor and open
File>Preferences>Cursor Settings, InCursor Settings, clickMCP, and make sureextend-sdk-mcp-serveris enabled.
[!IMPORTANT] Use the
ghcr.io/accelbyte/ags-extend-sdk-mcp-serverimage tag that matches your AGS version. See the available image tags here.
Sample prompts
In Cursor, press CTRL+L and try the following prompts. You should see that the tools provided by this MCP server are used. Give permission to execute the tools when requested.
- Search symbols:
Search for symbols related to 'user' - Get symbol details:
Describe the 'AdminCreateUser@iam' and 'User@iam' symbols
[!TIP] When coding using this MCP server, we recommend to start from an Extend SDK getting started sample project or an Extend app template project instead of a blank project. Add the necessary context, such as specific source code files, to help getting better results.
Environment Variables
TRANSPORT: The MCP server transport (valid values:stdio,http,streamableHttp, default:stdio)PORT: HTTP server port ifTRANSPORTishttp(default:3000)CONFIG_DIR: Directory of YAML config files (recursive, default:config/go)- For Extend SDK C#:
config/csharp - For Extend SDK Go:
config/go - For Extend SDK Java:
config/java - For Extend SDK Python:
config/python
- For Extend SDK C#:
LOG_LEVEL: Logging level (valid values:debug,info,warn,error, default:info)ALLOWED_ORIGINS: Comma-separated list of allowed origins for HTTP transport (optional)NODE_ENV: Environment (valid values:development,production) (optional, used by Express for HTTP transport)
Development
Prerequisites
- Bash
- Curl
- Docker
- Makefile
- Node.js 18+
- pnpm
Install dependencies
pnpm install
Start the MCP server for development
With the default STDIO transport
pnpm dev
With streamable HTTP transport
TRANSPORT=http pnpm dev
Build the MCP server
pnpm build
Start the MCP server after build
With the default STDIO transport
pnpm start
With streamable HTTP transport
TRANSPORT=http pnpm start
Build the MCP server container image
docker build -t extend-sdk-mcp-server:latest .
Release
Push the MCP server container image to container registry
# Setup variables
GHCR_USERNAME=<your-username>
GHCR_PASSWORD=<your-password>
IMAGE_TAG=2026.2.0 # Matches AGS release, bump patch version for hotfix
# Prepare builder
docker buildx inspect extend-sdk-mcp-server-builder || docker buildx create --name extend-sdk-mcp-server-builder --use
# Login, build, and push multiarch image
docker login --username ${GHCR_USERNAME --password $GHCR_PASSWORD}
docker buildx build -t ghcr.io/accelbyte/ags-extend-sdk-mcp-server:${IMAGE_TAG} --platform linux/amd64,linux/arm64 --push .
# Clean up builder
docker buildx rm --keep-state extend-sdk-mcp-server-builder
Testing
-
Start the MCP server with HTTP transport.
-
Initialize the MCP connection.
curl -N -H "Accept: application/json, text/event-stream" \ -H "Content-Type: application/json" \ -X POST \ -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"test-client","version":"1.0.0"}}}' \ http://localhost:3000/ -
List available tools.
curl -N -H "Accept: application/json, text/event-stream" \ -H "Content-Type: application/json" \ -X POST \ -d '{"jsonrpc":"2.0","id":2,"method":"tools/list","params":{}}' \ http://localhost:3000/ -
Test the search tool.
curl -N -H "Accept: application/json, text/event-stream" \ -H "Content-Type: application/json" \ -X POST \ -d '{"jsonrpc":"2.0","id":3,"method":"tools/call","params":{"name":"search-symbols","arguments":{"query":"user"}}}' \ http://localhost:3000/ -
Test describe model.
curl -N -H "Accept: application/json, text/event-stream" \ -H "Content-Type: application/json" \ -X POST \ -d '{"jsonrpc":"2.0","id":4,"method":"tools/call","params":{"name":"describe-symbols","arguments":{"ids":["User@iam"]}}}' \ http://localhost:3000/
Servidores relacionados
Scout Monitoring MCP
patrocinadorPut performance and error data directly in the hands of your AI assistant.
Alpha Vantage MCP Server
patrocinadorAccess financial market data: realtime & historical stock, ETF, options, forex, crypto, commodities, fundamentals, technical indicators, & more
Merge MCP Server
Integrates the Merge Unified API with any LLM provider using the MCP protocol.
MCP Aggregator
An MCP (Model Context Protocol) aggregator that allows you to combine multiple MCP servers into a single endpoint allowing to filter specific tools.
Remote MCP Server (Authless)
An example of a remote MCP server deployable on Cloudflare Workers, without authentication.
MCP迭代管理工具
An iteration management tool to automate the collection and submission of iteration information to a CodeReview system.
codeix
Fast semantic code search for AI agents — find symbols, references, and callers across any codebase. Pre-built index committed to git, instant queries via MCP.
Claude Code Memory Server
A Neo4j-based MCP server providing persistent memory and contextual assistance for Claude Code.
Sentry
Interact with the Sentry API to monitor application errors and performance.
better-code-review-graph
Knowledge graph for token-efficient code reviews with Tree-sitter parsing, dual-mode embedding (ONNX + LiteLLM), and blast-radius analysis via MCP tools.
Pprof Analyzer
Analyze Go pprof performance profiles (CPU, heap, goroutine, etc.) and generate flamegraphs.
MCP All Servers
A collection of reference implementations for the Model Context Protocol (MCP), showcasing servers built with TypeScript and Python SDKs.