transcend

transcend

A code-intelligence MCP server for coding agents that combines a static repository map (via tree-sitter) with precise live navigation (via LSP), providing structured, token-budgeted responses anchored to file:line.

Category
访问服务器

README

transcend

Docs: ldippo.github.io/transcend-mcp · Suite: transcend-harness

A code-intelligence MCP server for coding agents: a cheap static map of a repository (tree-sitter structural graph — clusters, hubs, budgeted subgraphs) plus precise live navigation (real language servers via LSP), bridged by stable symbol IDs. Every response is structured, token-budgeted, and anchored to file:line — never a raw file dump.

Languages: Python (pyright) and TypeScript/JavaScript (typescript-language-server). Adding a language = a tree-sitter grammar + .scm queries + an LSP command — no core changes.

Install & run

npm install
npm run build

# optional but recommended — without them the server runs map-only:
npm install -g pyright typescript-language-server typescript

node dist/src/index.js --root /path/to/repo        # stdio MCP server
node dist/src/index.js --root /path/to/repo --no-watch

On first run the server indexes the repo in the background (≈0.5s per 1k files) and persists the index under <repo>/.transcend/ (add it to .gitignore). A file watcher keeps the index fresh incrementally.

Wire into a harness

Any MCP-over-stdio harness works. Claude Code:

claude mcp add transcend -- node /path/to/transcend/dist/src/index.js --root /path/to/repo

Generic MCP config:

{
  "mcpServers": {
    "transcend": {
      "command": "node",
      "args": ["/path/to/transcend/dist/src/index.js", "--root", "/path/to/repo"]
    }
  }
}

Give the agent docs/AGENT_GUIDE.md — it documents the orient-with-map / confirm-with-LSP policy that the tool descriptions also encode.

Tool surface

Tool Layer Purpose
map_overview map clusters + hub symbols; start here
map_search map fuzzy symbol lookup → node IDs
map_neighbors map callers/callees/imports/inheritance around a node
map_path map shortest structural path between two symbols
map_rebuild, map_status map index management + both layers' health
nav_definition / nav_references / nav_implementations live ground truth, always current
nav_type / nav_symbols / nav_workspaceSymbols live hover, file outline, workspace search
nav_callHierarchy live incoming/outgoing calls (capability-gated, with fallback)
resolve bridge node ID ⇄ verified live file:line:col, staleness flagged

All tools accept tokenBudget (default 2000, max 10000) and return a uniform envelope; see the agent guide.

Development

npm test          # unit + integration (LSP tests auto-skip if servers absent)
npm run smoke     # build + scripted MCP client against the TS fixture
npx @modelcontextprotocol/inspector node dist/src/index.js --root test/fixtures/ts-mini

Architecture notes: docs/ARCHITECTURE.md.

No telemetry. No network calls. Secrets are never read; configuration is CLI flags only.

推荐服务器

Baidu Map

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。

官方
精选
JavaScript
Playwright MCP Server

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。

官方
精选
TypeScript
Magic Component Platform (MCP)

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。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

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。

官方
精选
本地
TypeScript
VeyraX

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。

官方
精选
本地
graphlit-mcp-server

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。

官方
精选
TypeScript
Kagi MCP Server

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。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

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。

官方
精选