Aider RepoMapper MCP Server
Generates optimized repository maps and enables identifier search with graph-ranking, designed for efficient LLM context caching and codebase navigation.
README
Aider RepoMapper MCP Server
A standalone Model Context Protocol (MCP) server wrapping Aider's graph-ranking repository mapping engine. This server is explicitly hardened for high-efficiency LLM development workflows by optimizing text payloads for strict byte-for-byte Prompt Caching (e.g., Anthropic Prompt Caching via OpenRouter).
Key Features
- Cache-Locked Architecture Summarization: Intercepts volatile file parsing outputs and swaps mutating line numbers for static pipe (
|) tokens. - Rank Score Stripping: Removes fluctuating PageRank floating-point scores (
(Rank value: X.XXXX)) generated dynamically in memory, preventing prefix cache invalidation loops across code changes. - Path Blacklisting Heuristics: Automatically intercepts and filters untracked runtime assets, database dumps, localized state subtrees (
data/), and compiled frontend asset bundles (*.min.js) from corrupting the tool context window. - JSON-RPC Schema Compliant: Plugs natively into agent frameworks like ZooCode, RooCode, and Cursor-agent extension environments out of the box.
Server Layout & Core Modules
The repository structures the port across these primary utilities:
repomap_server.py: Async JSON-RPC server harness exposing tool schemas and managing filesystem parameter inputs.repomap_class.py&repomap.py: Local interface adapters wrapping the core Abstract Syntax Tree (AST) tree-sitter parser hooks.importance.py&scm.py: Heuristic logic for assessing file contribution ranks and structural tracking states.utils.py: Local counting frameworks and atomic file reader fallbacks.test_matrix.py: Regression framework validating formatting state guarantees.
Installation & Environment Configuration
Ensure the engine is running within a Python 3.10+ environment containing the core dependency stack (mcp, tiktoken, aider-chat).
Agent Settings Integration
Add the server entry point configuration to your global or project-level MCP definition matrix (e.g., .vscode/mcp.json or .roo/mcp.json):
{
"mcpServers": {
"repo-mapper": {
"command": "/home/barry/tools/RepoMapper/.venv/bin/python",
"args": ["/home/barry/tools/RepoMapper/repomap_server.py"],
"env": {
"PYTHONUNBUFFERED": "1"
}
}
}
}
Exposed MCP Tools
repo_map
Generates an optimized, topologically sorted layout matching the internal repository architecture.
Parameters:
project_root(string, required): The absolute path to the targeted repository root.token_limit(integer, optional): Maximum scale window for the map text block (defaults to 2000).exclude_unranked(boolean, optional): Drops low-signal definitions when processing constrained tokens.verbose(boolean, optional): Promotes deeper diagnostic stdout updates to log levels.
search_identifiers
Performs keyword-based target cross-referencing against structural symbol matrices.
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。