mcp-server-ts-analysis
MCP server for TypeScript type resolution and dependency graph analysis using ts-morph and madge.
README
mcp-server-ts-analysis
MCP server for TypeScript type resolution and dependency graph analysis. Uses ts-morph for type intelligence and madge for dependency graphs. Stdio transport.
Setup
1. Clone and build
git clone https://github.com/akshay-nm/mcp-server-ts-analysis.git
cd mcp-server-ts-analysis
npm install
npm run build
2. Add to Claude Code
Global (all projects) — add to ~/.claude.json:
{
"mcpServers": {
"ts-analysis": {
"type": "stdio",
"command": "node",
"args": ["/absolute/path/to/mcp-server-ts-analysis/dist/bin/cli.js"]
}
}
}
Per project — add to .mcp.json in the project root:
{
"mcpServers": {
"ts-analysis": {
"type": "stdio",
"command": "node",
"args": ["./path/to/mcp-server-ts-analysis/dist/bin/cli.js"]
}
}
}
3. Restart Claude Code
The server will appear in your MCP server list. No startup config needed — all paths (tsconfig, source_root, etc.) are passed per tool call, so one server instance works across all your projects.
Other MCP clients
Any MCP client that supports stdio transport can use this server. Point it at dist/bin/cli.js with Node.js as the command.
Tools
Type analysis
| Tool | Description |
|---|---|
resolve_type |
Fully computed type at a source position |
hover_info |
Quick info similar to VS Code hover (type, docs, JSDoc tags) |
type_diagnostics |
TypeScript errors and warnings, optionally scoped to a file |
Common parameters:
file— absolute path to the TypeScript fileline— line number (1-based)col— column number (1-based)tsconfig— path to tsconfig.json (optional, uses default compiler options if omitted)
Dependency analysis
| Tool | Description |
|---|---|
dep_graph |
Full dependency tree as JSON |
reverse_deps |
All files that import a given file |
forward_deps |
All files that a given file imports |
circular_deps |
All circular dependency chains |
import_path |
Shortest import chain between two files |
Common parameters:
source_root— source root directory to analyze (required)tsconfig— path to tsconfig.json (optional)exclude— directories to exclude (defaults tonode_modules,dist,.git)exclude_patterns— regex patterns to exclude files (e.g.["\\.d\\.ts$"]to skip declaration files)
dep_graph extras:
entry— entry file to start traversal from (defaults to auto-detected root nodes)max_depth— max traversal depth from entry/root nodes (omit for full graph)
Examples
Resolve the type of a variable at line 10, column 7:
{
"tool": "resolve_type",
"args": {
"file": "/home/user/project/src/server.ts",
"line": 10,
"col": 7,
"tsconfig": "/home/user/project/tsconfig.json"
}
}
Get the top-level module boundaries (depth 1) without .d.ts files:
{
"tool": "dep_graph",
"args": {
"source_root": "/home/user/project/src",
"tsconfig": "/home/user/project/tsconfig.json",
"exclude_patterns": ["\\.d\\.ts$"],
"max_depth": 1
}
}
Find how auth.ts reaches database.ts through imports:
{
"tool": "import_path",
"args": {
"fileA": "auth.ts",
"fileB": "database.ts",
"source_root": "/home/user/project/src"
}
}
Architecture
src/
index.ts — MCP server, tool registration, stdio transport
type-analysis.ts — ts-morph type resolution (cached by tsconfig path)
dep-analysis.ts — madge dependency graph analysis
types.ts — shared interfaces
madge.d.ts — type declarations for madge
bin/
cli.ts — entry point
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。