发现优秀的 MCP 服务器

一个基于Blender-mcp的研究项目，自学记录用( Blender-mcp详情见

好的，我将尽力帮助你理解 Blender-mcp 项目，判断其能力边界，理解 mcp 架构，并尝试添加新功能。 首先，我们需要明确几个步骤： 1. **理解 Blender-mcp 项目:** * **项目目标:** 了解该项目旨在解决什么问题，例如：多客户端协作、版本控制、资源管理等。 * **核心功能:** 掌握项目已经实现的功能，例如：场景同步、资产共享、实时聊天等。 * **技术栈:** 了解项目使用的编程语言、框架、库等，例如：Python、Blender API、网络协议等。 * **代码结构:** 熟悉项目的代码组织方式，例如：客户端、服务器端、数据模型等。 2. **判断能力边界:** * **现有功能的局限性:** 思考现有功能在哪些方面存在不足，例如：同步速度、稳定性、可扩展性等。 * **不支持的功能:** 识别项目尚未实现，但可能对用户有价值的功能，例如：高级冲突解决、权限管理、插件支持等。 * **性能瓶颈:** 分析项目在处理大规模场景或高并发请求时可能遇到的性能问题。 3. **理解 mcp 架构:** * **架构模式:** 确定项目采用的架构模式，例如：客户端-服务器、P2P、混合模式等。 * **组件划分:** 了解项目由哪些核心组件构成，例如：同步引擎、数据存储、通信模块等。 * **数据流:** 跟踪数据在不同组件之间的流动方式，例如：场景数据如何从客户端同步到服务器端。 * **扩展点:** 识别架构中可以方便地添加新功能的扩展点，例如：插件接口、自定义数据类型等。 4. **尝试添加新功能:** * **选择功能:** 基于对项目和架构的理解，选择一个具有可行性和价值的新功能。 * **设计方案:** 设计新功能的实现方案，包括：数据结构、算法、用户界面等。 * **编码实现:** 编写代码实现新功能，并进行测试和调试。 * **集成测试:** 将新功能集成到现有项目中，并进行全面的测试。 **具体步骤和建议:** * **阅读项目文档:** 仔细阅读 Blender-mcp 项目的 README 文件、API 文档、设计文档等，了解项目的背景、目标、架构和使用方法。 * **运行项目:** 尝试运行 Blender-mcp 项目，体验其现有功能，并观察其运行状态。 * **阅读源代码:** 阅读 Blender-mcp 项目的源代码，理解其实现细节，并学习其编程风格。 * **调试项目:** 使用调试器调试 Blender-mcp 项目，跟踪代码的执行流程，并查找潜在的问题。 * **参与社区:** 加入 Blender-mcp 项目的社区，与其他开发者交流，并寻求帮助。 * **从小处着手:** 从简单的功能开始，逐步增加复杂性，避免一开始就尝试修改核心架构。 * **编写测试用例:** 为新功能编写测试用例，确保其正确性和稳定性。 * **提交代码:** 将新功能的代码提交到 Blender-mcp 项目的仓库，并接受代码审查。 **关于添加新功能的建议:** * **冲突解决:** 改进现有的冲突解决机制，例如：支持更细粒度的冲突检测、提供更友好的冲突解决界面。 * **权限管理:** 添加权限管理功能，允许用户控制其他用户对场景的访问和修改权限。 * **版本控制:** 集成版本控制系统，例如：Git，允许用户回滚到之前的版本。 * **插件支持:** 提供插件接口，允许开发者扩展 Blender-mcp 的功能。 * **性能优化:** 优化同步算法，减少网络延迟，提高同步速度。 **翻译:** 通过学习 Blender-mcp 项目，判断其能力边界，并理解 mcp 架构，尝试添加新的功能。 通过学习 Blender-mcp 项目，判断其能力边界，并理解 mcp 架构，尝试添加新的功能。 **中文翻译:** 通过学习 Blender-mcp 项目，判断其能力边界，并理解 mcp 架构，尝试添加新的功能。 **更自然的中文翻译:** 通过研究 Blender-mcp 项目，确定它的能力范围，理解其 mcp 架构，并尝试添加新的功能。 希望这些信息能帮助你更好地理解 Blender-mcp 项目，并成功添加新功能。祝你一切顺利！

MCP Empe Server

Enables users to manage self-sovereign identity credentials by creating schemas, issuing verifiable credentials, generating verification QR codes, and checking verification status via the Empe SSI ecosystem.

renef-mcp

MCP server for Renef, providing dynamic instrumentation and reverse engineering capabilities for Android ARM64, including native hooking, Java interception, and memory manipulation.

Unified Agent Context System

A highly configurable MCP server with modular architecture supporting portfolio management and test tools over HTTP or stdio transport.

Sleeper Fantasy MCP

Enables comprehensive Sleeper Fantasy Football integration with Claude, providing real-time player projections, historical performance analytics, league management, and waiver wire analysis. Supports advanced NFL metrics, lineup optimization, and matchup analysis for fantasy football decision-making.

Screener MCP Server

MCP server for Indian stock market data — search companies, analyze fundamentals, compare stocks, browse sectors and screens.

hive-mcp-openclaw-bridge

MCP server that wraps Hive Gamification's 8 BOGO doors behind a 4-level hierarchical budget with atomic reservation, upstream POST, and settlement/release, implementing OpenClaw's 3-layer payment governance.

jlc-mcp

Local Model Context Protocol server that exposes EasyEDA Pro design capabilities as MCP tools, enabling project inspection, schematic editing, PCB design, validation, and manufacturing exports through natural language.

Proflores MCP Server

Enables management of the Proflores budgeting system, allowing users to interact with catalogs for clients, projects, and resources. It supports comprehensive budget operations including creation, item updates, and Unit Price Analysis (APU) through a Supabase backend.

Unlock-MCP

An MCP server providing comprehensive access to Unlock Protocol on Base and Base-Sepolia networks, enabling AI agents to interact with membership, subscription, and access control functionality.

epwforge-mcp

Weather files (EPW/DDY) for building energy simulation, on-demand. Four tools: search the 17,000-station catalog, analyze any EPW (design conditions, HDD/CDD, DesignDay IDF emit), render 5 chart types (diurnal, wind rose, temp carpet, monthly box plot, comparison), and generate TMY/AMY/CMIP6-morphed scenarios with UHI, extreme events, and wildfire smoke overlays. 3 of 4 tools work without API key.

posthog-mcp

Enables interaction with PostHog product analytics API, supporting project listing, event queries, insight analysis, feature flag management, and annotation creation.

GithubMCP

Connects AI assistants to GitHub repositories, pull requests, issues, commits, and code search while enabling repository visibility controls, CI/CD monitoring, sandboxed local filesystem access, and code quality/security analysis.

MCP AI Agent Server

Bridges Cline to an AI agent system through the Model Context Protocol, enabling message sending, session management, and assistant switching for seamless interaction with AI agents.

android-mcp-server

Enables Android device control via ADB and scrcpy, including device management, app management, file operations, shell commands, input simulation, screen mirroring, and UI element inspection.

kie-image-mcp

Enables image generation using KIE.ai models like google/nano-banana and GPT Image 2. Supports asynchronous single or parallel image generation and file uploading for image inputs.

Smithery Registry OpenAPI Server

用于 Smithery Registry 的 OpenAPI 服务器，提供对 MCP 服务器的访问

Kickd MCP Server

Enables Claude Code to manage automations, tasks, skills, workflows, credentials, variables, webhooks, events, and notifications within the kickd background automation daemon.

API to MCP

Converts any REST API endpoints into MCP tools, enabling AI clients like Cursor and Claude Desktop to call internal services directly.

Slander MCP

An MCP server that retrieves and ranks humorous roasts, jokes, and memes about real or fictional characters from Twitter/X. It leverages LLMs for intelligent query generation and uses social engagement metrics to identify and rank high-quality content.

instagram-analyzer MCP

Analyzes public Instagram post URLs, returning captions and photos as image content for multimodal models like Claude, without requiring paid vision APIs.

Pytheum MCP

Wraps the four public endpoints on api.pytheum.com as MCP tools so AI agents can ask which markets a news article moves without writing their own retrieval layer.

Fitbit Air Daily Planner MCP Server

MCP server that retrieves health data from Google Health API and calendar events from Google Calendar, and enables ChatGPT to create personalized wellness plans that are written to a dedicated calendar only upon user approval.

TrendRadar

TrendRadar aggregates hot news from multiple platforms and provides AI-powered trend analysis via the Model Context Protocol. It enables users to filter for relevant information and receive automated updates across various notification channels like Telegram, Slack, and WeChat.

MemoryMesh

这个项目基于 MCP 服务器仓库中的知识图谱内存服务器，并保留了其核心功能。

tossinvest-api-mcp

A local MCP server that automatically generates tools from Toss Securities' official OpenAPI spec, enabling real API calls with multi-layered order safety and OAuth 2.0 authentication.

MongoDB MCP Server

好的，让我来帮你翻译成中文： **让我们创建一个 MongoDB MCP 服务器。** (Ràng wǒmen chuàngjiàn yī ge MongoDB MCP fúwùqì.)

Databricks MCP Server

一个实现了模型补全协议（MCP）的服务器，它允许大型语言模型（LLM）通过自然语言与 Databricks 资源进行交互，包括集群、作业、笔记本和 SQL 执行。

dat-ai-mcp

Provides browser automation, audio transcription, and LLM chat as MCP tools for any agent.