Best AI Repos & MCP Servers for Content Creators
Top open-source AI repositories, MCP servers, and creator tools. Curated for content creators building with AI.
Repositories
langgenius/dify
Dify is an open-source platform for building production-grade AI applications through a visual drag-and-drop workflow builder. Instead of writing boilerplate code to chain LLM calls, manage prompts, and wire up retrieval pipelines, developers lay out their logic on a canvas -- connecting model nodes, tool calls, conditional branches, and human-in-the-loop checkpoints into executable workflows. The result is a system that can go from prototype to production deployment without rewriting the orchestration layer. The platform supports hundreds of LLM providers out of the box: OpenAI GPT models, Anthropic Claude, Mistral, Llama 3, Qwen, and any provider exposing an OpenAI-compatible API. Switching between models is a dropdown change, not a code refactor. This provider-agnostic design means teams can start with a cloud API, benchmark against alternatives, and migrate to self-hosted models without touching their workflow logic. Dify ships with a full RAG pipeline built in. Upload PDFs, presentations, or plain text, and the system handles chunking, embedding, vector storage, and retrieval. Version 1.12.0 introduced Summary Index, which attaches AI-generated summaries to document chunks so semantically related content clusters together during retrieval. Version 1.13.0 added multimodal retrieval that unifies text and images into a single semantic space for vision-enabled reasoning. The agent capabilities layer supports both function-calling and ReAct-style reasoning with over 50 built-in tools including Google Search, DALL-E, Stable Diffusion, and WolframAlpha. Since v1.0.0, all models and tools have been migrated to a plugin architecture, so extending Dify with custom integrations no longer requires forking the core codebase. On the operational side, Dify includes a Prompt IDE for comparing model outputs side-by-side, LLMOps-grade logging and performance monitoring, and a Backend-as-a-Service API layer that lets frontend applications consume workflows through REST endpoints. The v1.13.0 release added a Human Input node that pauses workflows for human review, enabling approval gates and content moderation loops directly within automated pipelines. Deployment is flexible: Docker Compose for quick self-hosted setups (minimum 2 CPU cores, 4 GB RAM), Kubernetes via five community-maintained Helm charts, Terraform templates for Azure and Google Cloud, and AWS CDK for infrastructure-as-code deployments. Dify Cloud offers a managed option with 200 free GPT-4 calls to get started. Enterprise customers including Volvo Cars and Kakaku.com run Dify in production -- Kakaku.com reported 75% employee adoption with nearly 950 internal AI applications built on the platform.
132,700 starsanomalyco/opencode
118,000 stars in under a year. That's not a typo. OpenCode went from zero to the most-starred AI coding agent on GitHub faster than any developer tool in memory -- and it did it by being the one thing Cursor and Claude Code refuse to be: completely open source, completely provider-agnostic, and completely free. Here's the pitch that mass-converted developers: bring your own model, bring your own keys, keep your own data. OpenCode doesn't care if you're running Claude Opus, GPT-4.1, Gemini 2.5, or a local Llama instance through Ollama. It treats every provider as a first-class citizen. If you've ever felt locked into Anthropic's pricing because Claude Code only works with Claude, or locked into Cursor's $20/month because switching means losing your workflow -- OpenCode is the exit door 5 million developers already walked through. The architecture is what makes it stick. OpenCode runs a client/server split -- the AI agent runs as a background server while the TUI, desktop app, or IDE extension connects as a client. That means you can run the agent on a beefy remote machine and code from a thin laptop over SSH. Try doing that with Cursor. Two built-in agents handle different workflows: a build agent for writing and modifying code, and a read-only plan agent for exploring codebases without accidentally changing anything. There's also a general subagent that handles complex multi-step searches. LSP integration gives it real code intelligence -- not just pattern matching, but actual type-aware navigation and diagnostics. The release velocity tells its own story: 731 releases, 10,045 commits, v1.2.21 shipped March 7, 2026. The team at Anomaly (the same folks behind terminal.shop) ships daily. MCP support means you can extend it with the same server ecosystem Claude Code uses. Install takes one curl command. You're coding in 30 seconds.
121,683 starsinfiniflow/ragflow
RAGFlow is a leading open-source Retrieval-Augmented Generation engine that fuses deep document understanding with agentic AI capabilities to build a superior context layer for large language models. Unlike general-purpose RAG frameworks, RAGFlow specializes in extracting structured knowledge from complex, visually rich documents — including PDFs with tables, multi-column layouts, images, scanned copies, spreadsheets, slides, and web pages — with high fidelity. The platform provides template-based intelligent chunking with visual customization, high-precision hybrid search combining vector search, BM25, and custom scoring with advanced re-ranking, and grounded citations that reduce hallucinations by linking every answer back to traceable source references. RAGFlow includes a visual workflow builder for designing agentic RAG pipelines with memory support, Model Context Protocol (MCP) integration, and multi-modal model support for processing images within documents. It ships with Docker-based deployment in both lightweight (2 GB) and full-featured (9 GB) configurations, supports Elasticsearch and Infinity as storage backends, and works with configurable LLMs and embedding models. With 74,000+ GitHub stars and an Apache 2.0 license, RAGFlow has become one of the most popular open-source RAG solutions, particularly for enterprise use cases in equity research, legal analysis, and manufacturing where document intelligence is critical.
74,955 starsmsitarzewski/agency-agents
Agency-Agents is a production-ready collection of 144+ specialized AI agent personas organized across 12 divisions — Engineering, Design, Paid Media, Sales, Marketing, Product, Project Management, Testing, Support, Spatial Computing, Game Development, and Specialized. Each agent is a structured Markdown file that gives any LLM a specific professional identity, complete with domain expertise, a distinct communication style, battle-tested workflows, concrete deliverables, and measurable success metrics. Rather than relying on generic prompting, you activate a focused expert — a Frontend Wizard, Security Engineer, Brand Guardian, UX Researcher, or one of 140+ others — and the LLM narrows its context accordingly, reducing hallucinations and enforcing domain best practices. The project was born from a Reddit discussion about AI agent specialization and grew through months of community iteration into one of the fastest-starred repositories on GitHub. It integrates natively with Claude Code's /agents system by placing Markdown files in ~/.claude/agents/, and ships automated install scripts that convert agents for Cursor, Aider, Windsurf, Gemini CLI, OpenCode, and more. With 43.9K stars and 6.6K forks as of March 2026, agency-agents has become the de facto starting point for teams that want to run structured multi-agent workflows from their IDE.
43,880 starsItzCrazyKns/Perplexica
Perplexica is a privacy-focused, open-source AI-powered answering engine designed to run entirely on your own hardware. Often described as an open-source alternative to Perplexity AI, Perplexica combines real-time internet search with the intelligence of large language models to deliver accurate, cited answers to complex queries without compromising user privacy. At its core, Perplexica leverages SearxNG as its search backbone, a privacy-respecting metasearch engine that aggregates results from multiple sources without tracking users. The retrieved results are then processed through a Retrieval-Augmented Generation (RAG) pipeline, where an LLM synthesizes the information into coherent, source-cited responses. This architecture ensures that every answer is grounded in verifiable web content rather than relying solely on the model's training data. One of Perplexica's most compelling features is its multi-model flexibility. Users can connect to virtually any LLM provider, including OpenAI, Anthropic Claude, Google Gemini, Groq, and locally hosted models through Ollama. This means developers and privacy-conscious users can run the entire stack on-premises with no data leaving their network, or mix cloud and local models depending on the task. Perplexica offers three distinct search modes tailored to different needs. Speed Mode prioritizes quick answers for simple lookups. Balanced Mode handles everyday research with a good tradeoff between depth and response time. Quality Mode performs deep, multi-step research for thorough investigation of complex topics. Beyond web search, the engine supports academic paper search, discussion forum search, image and video search, and domain-restricted queries. The platform also includes smart contextual widgets that surface relevant quick-lookup information such as weather forecasts, mathematical calculations, and stock prices directly in the search interface. Users can upload files including PDFs, text documents, and images for the AI to analyze alongside web results. All search history is saved locally, giving users full control over their data. With over 31,000 GitHub stars, 3,300 forks, and 44 contributors, Perplexica has established itself as one of the most popular open-source AI search projects. The project ships with Docker support for easy deployment, including a bundled SearxNG option that gets everything running with a single command. One-click deployment is also available through platforms like Sealos, RepoCloud, and Hostinger. A developer-facing API allows integration of Perplexica's search capabilities into custom applications and workflows.
32,909 starsQwenLM/Qwen3
Qwen3 is the flagship open-weight large language model series from Alibaba Cloud's Qwen team, offering one of the most comprehensive lineups in the open-source AI ecosystem. The repository serves as the central hub for a family of models spanning dense architectures (0.6B, 1.7B, 4B, 8B, 14B, and 32B parameters) and mixture-of-experts designs (30B-A3B and 235B-A22B), giving developers and researchers granular control over the compute-performance tradeoff for their specific deployment scenario. What distinguishes Qwen3 from other open-weight model families is its hybrid thinking architecture. Every model in the series supports seamless switching between a step-by-step reasoning mode for complex logic, mathematics, and code generation, and a rapid non-thinking mode for straightforward queries. Users can configure thinking budgets to balance latency against reasoning depth, making the models adaptable to both real-time applications and offline batch processing. Trained on approximately 36 trillion tokens, double the training corpus of Qwen2.5, the models demonstrate strong multilingual capabilities across 119 languages and dialects. Context windows range from 32K tokens on smaller models to 128K on larger variants, with experimental support extending to 1 million tokens in the Qwen3-2507 update released in August 2025. The series has evolved further with Qwen3.5, which introduced compact models from 0.8B to 9B parameters optimized for on-device deployment using a hybrid Gated Delta Networks and sparse MoE architecture. Qwen3 integrates natively with popular inference frameworks including vLLM, SGLang, TensorRT-LLM, llama.cpp, and Ollama, and ships with enhanced agentic capabilities for tool calling and MCP (Model Context Protocol) support. All models are released under the Apache 2.0 license and available on Hugging Face, ModelScope, and Kaggle.
26,896 starslightpanda-io/browser
Lightpanda is the headless browser built from scratch for AI agents and automation. While Puppeteer and Playwright run Chrome under the hood — a 400 MB browser that takes 2+ seconds to cold-start — Lightpanda launches in milliseconds and uses 9x less memory with 11x faster execution. It speaks the same Chrome DevTools Protocol (CDP) so Playwright, Puppeteer, and raw CDP clients all work without code changes. Built in Zig with a custom JavaScript engine via v8, Lightpanda handles full DOM APIs, Ajax (XHR and Fetch), and network interception. It's not a simple HTML scraper. It actually runs JavaScript, renders dynamic content, and supports modern web standards — just without the overhead of a full desktop browser. For AI agents doing web research, data extraction, or automated testing at scale, the performance gap is significant: one cloud instance can run 11 Lightpanda sessions where Chrome would run one. The project hit 17,000+ GitHub stars with over 2,000 new stars in a single day, driven by growing demand for AI-native infrastructure tools. An official MCP server (gomcp) enables direct integration with Claude Code, Cursor, and other AI coding agents through the Model Context Protocol. The browser is available as nightly binaries for Linux x86_64 and macOS aarch64, plus Docker images for both architectures. GPL-2.0 licensed with active development.
17,087 starspromptfoo/promptfoo
Promptfoo is a CLI and library that eliminates trial-and-error from LLM application development. Instead of manually testing prompts and hoping they work, you define test cases with expected outputs, run them against multiple models simultaneously, and get a pass/fail report with a visual dashboard. The tool supports every major provider — OpenAI, Anthropic, Azure, Bedrock, Ollama, and dozens more — so you can compare model performance side-by-side without rewriting code. Define your evaluations in YAML, run them from the terminal, and view results in a browser-based comparison UI. What sets promptfoo apart from other eval frameworks is its red-teaming capability. Beyond functional testing, it scans your LLM apps for security vulnerabilities: prompt injection, jailbreaks, PII leakage, and harmful content generation. This makes it both a quality assurance tool and a security scanner in one package. The developer experience is polished. Local-first execution means your data never leaves your machine. Built-in caching speeds up repeated runs. CI/CD integration lets you block deployments when prompt quality drops. And the PR review feature automatically flags LLM security issues in pull requests. With 16.4K GitHub stars, 398 releases, and active maintenance (latest release March 12, 2026), promptfoo has become the de facto standard for teams that take LLM output quality seriously.
16,408 starsVoltAgent/awesome-agent-skills
A curated directory of 549+ agent skills built by real engineering teams at Anthropic, Google Labs, Vercel, Stripe, Cloudflare, Netlify, Sentry, Trail of Bits, Hugging Face, Expo, Microsoft, and dozens of other organizations. Unlike the typical awesome-list that dumps links without context, this repository organizes skills by provider with expandable detail sections, direct GitHub links to each skill implementation, and clear compatibility notes for the AI coding assistants that actually support them: Claude Code, Codex, Antigravity, Gemini CLI, Cursor, GitHub Copilot, OpenCode, and Windsurf. The skill categories span the full range of developer workflows. You get official Claude skills for document generation, frontend design, canvas artifacts, and MCP server building. Composio alone contributes integrations with 1000+ external apps. Stripe, Supabase, HashiCorp, Sanity, Neon, ClickHouse, Remotion, and Replicate each maintain their own dedicated skill sets. Trail of Bits provides security-focused agent skills for auditing. Language-specific sections cover .NET, Java, Python, Rust, and TypeScript, while general-purpose skills handle marketing, productivity, testing, and n8n workflow automation. The repository carries an important security disclaimer: all skills are curated, not audited. Maintainers can update or replace skills at any time, and VoltAgent recommends running the Synk Skill Security Scanner before installing anything into your development environment. With 9.5K GitHub stars, 817 forks, and 226 commits on main, this is the most actively maintained index of agent skills available anywhere.
11,156 starsvikhyat/moondream
Moondream is an open-source family of vision language models (VLMs) engineered for powerful, efficient visual reasoning at a fraction of the size of competing models. The latest release, Moondream 3 Preview, uses a mixture-of-experts architecture with 9B total parameters but only 2B active during inference, delivering state-of-the-art results in object detection (88.6% on RefCOCOg), counting (93.2% on CountbenchQA), document understanding (86.6% on ChartQA), and hallucination resistance (89.0% on POPE) while fitting comfortably on edge hardware. Four built-in vision skills -- object detection, pointing and counting, visual question answering, and captioning -- cover the most common image understanding tasks out of the box. Moondream supports a 32K context window, grounded step-by-step reasoning that ties answers to spatial positions in an image, and a superword tokenizer that speeds text generation by 20-40%. Deployment is flexible: run locally via the free open-source Moondream Station, call the managed Moondream Cloud API, or self-host through platforms like Ollama and Hugging Face. With 3.5 million monthly downloads and adoption across retail, logistics, healthcare, and defense, Moondream has proven itself as the go-to lightweight VLM for production workloads ranging from media asset tagging and robotic vision to UI test automation.
9,418 starsalibaba/OpenSandbox
Running AI-generated code on your own machine without isolation is asking for trouble, and every developer building agents has felt that tension. OpenSandbox, open-sourced by Alibaba in March 2026, directly addresses this by giving you a production-grade sandbox platform purpose-built for AI workloads. It is the same infrastructure Alibaba uses internally for large-scale AI execution, now available under Apache 2.0. The architecture is clean and well-separated. A FastAPI-based lifecycle server manages sandbox creation and teardown through Docker or Kubernetes runtimes. Inside each isolated container, a high-performance Go-based execution daemon called execd handles command execution, filesystem operations, and code interpretation via internal Jupyter kernels. Communication across the stack is standardized through OpenAPI specifications, which means you can extend or replace components without rewriting the integration layer. What stands out is the breadth of SDK support. Python, Java, Kotlin, JavaScript, TypeScript, and C#/.NET clients are all available today, with Go on the roadmap. That level of polyglot coverage is rare for infrastructure projects at this stage. Each SDK wraps the unified sandbox API, so switching languages does not mean learning a new interface. On the security side, OpenSandbox supports three isolation tiers: gVisor for lightweight kernel-level sandboxing, Kata Containers for hardware-enforced VM isolation, and Firecracker microVMs for the strongest possible boundary with sub-second boot times. You pick the isolation level that matches your threat model rather than being locked into a single approach. The built-in environments cover the most common agent scenarios out of the box. You get command execution, filesystem access, and code interpretation as baseline capabilities. For GUI agents, there are pre-built browser automation setups with Chrome and Playwright, plus desktop environments accessible via VNC or VS Code Server. The network layer includes a unified ingress gateway with multiple routing strategies and per-sandbox egress controls, so you can restrict what each sandbox can reach on the internet. Getting started locally takes two commands: install via uv pip and run the init-config script. For production, the Kubernetes runtime enables distributed scheduling across clusters. The project hit 6,500 stars within days of release, with 468 forks and 628 commits already on main, signaling serious community traction and active development from Alibaba's team.
7,747 starszai-org/GLM-4.5
Finally, an open-source model that can actually use tools without falling apart. GLM-4.5 is Zhipu AI's (Z.ai) flagship Mixture-of-Experts foundation model built from the ground up for agentic workloads -- and it nails tool calling at a 90.6% success rate, beating even Claude Sonnet 4 (89.5%). The architecture packs 355 billion total parameters but only activates 32 billion per inference pass, making it roughly 8x more efficient than an equivalent dense model. That means you can run serious reasoning workloads without burning through your entire GPU budget. What makes GLM-4.5 genuinely interesting is the dual-mode design. Flip it into thinking mode for complex multi-step reasoning and tool orchestration, or run non-thinking mode when you just need a fast, direct response. This is not a gimmick -- on AIME 2024 math competition problems, thinking mode scores 91.0%, which blows past Claude Opus 4's 75.7%. The smaller GLM-4.5-Air variant (106B total, 12B active) still hits 89.4% on the same benchmark, which is absurd for a model you can run on just two H200 GPUs. The training pipeline is worth studying: Zhipu pretrained on 22 trillion tokens (15T text + 7T code/reasoning), then made three specialized copies of the base model -- one for reasoning, one for agentic tasks, one for general knowledge -- and distilled them back into a single unified model. The result is a model that scores 63.2 across 12 industry benchmarks, ranking 3rd globally behind only GPT-4 and Claude 4. Deployment is straightforward with SGLang or vLLM. FP8 quantization cuts hardware requirements in half (8x H100 instead of 16x), and the 128K context window handles long document workflows without chunking headaches. Everything ships under MIT license -- full commercial use, no restrictions, no catch. Weights are on both Hugging Face (43K+ monthly downloads) and ModelScope. The repo itself has 4.2K stars and includes inference code, deployment guides for Ascend NPUs and AMD GPUs, plus fine-tuning recipes for LLaMA-Factory and SWIFT.
4,265 starsMCP Servers
MCP Filesystem Server
The official Model Context Protocol filesystem server that gives Claude direct read and write access to files and directories on your local machine or server. Install it once and Claude can open files, create new ones, list directories, search across your codebase, and move or rename files — all without copy-pasting content into the chat window. It ships as part of the official MCP servers repository maintained by Anthropic, making it the most trusted and widely-deployed server in the ecosystem. Because you specify allowed directories at startup, Claude only ever sees what you explicitly permit — it cannot access /etc, your home folder, or any path outside the scope you configure.
81,124 starsMemory
The Memory MCP Server is an official Model Context Protocol reference server maintained by Anthropic that gives AI assistants persistent memory across conversations using a local knowledge graph. Without persistent memory, every conversation with an AI assistant starts from zero -- the model has no recollection of prior interactions, user preferences, or established context. The Memory server solves this by maintaining a structured knowledge graph stored as a local JSONL file that the assistant can read from and write to during any session. The knowledge graph is built on three core primitives. Entities are the primary nodes representing people, organizations, projects, concepts, or any other named object. Each entity has a unique name, a type classification, and a collection of observations. Relations are directed, labeled connections between entities stored in active voice (for example, "John_Smith works_at Anthropic"). Observations are discrete, atomic pieces of information attached to entities -- individual facts that can be independently added or removed without affecting other stored knowledge. The server exposes nine MCP tools that provide full CRUD operations over the knowledge graph: create_entities and create_relations for building the graph, add_observations for appending new facts to existing entities, delete_entities, delete_relations, and delete_observations for pruning outdated information, read_graph for retrieving the complete graph structure, search_nodes for querying entities by name, type, or observation content, and open_nodes for retrieving specific entities with their full context and interconnections. Storage uses a simple JSONL file format, making the knowledge base human-readable, version-controllable, and trivially portable. The storage path defaults to memory.jsonl in the working directory but can be customized via the MEMORY_FILE_PATH environment variable. Installation is straightforward through npx or Docker, and the server integrates natively with Claude Desktop, Claude Code, VS Code, Cursor, and other MCP-compatible clients. With over 44,000 weekly npm downloads and backing from the 80,000-star modelcontextprotocol/servers repository, the Memory server has become one of the most widely adopted MCP servers in the ecosystem.
81,024 starsPlaywright MCP Server
The Playwright MCP Server is Microsoft's official Model Context Protocol server that brings full browser automation capabilities to AI assistants and LLM-powered applications. Unlike traditional screenshot-based approaches that require vision models, this server leverages Playwright's accessibility tree to provide structured, deterministic representations of web page content that any LLM can process efficiently. This fundamental design choice makes it faster, more reliable, and more token-efficient than pixel-based alternatives. The server enables AI agents to navigate websites, click elements, fill forms, extract content, take screenshots, execute JavaScript, and interact with web applications through a comprehensive set of MCP tools. It supports all major browser engines including Chromium, Firefox, WebKit, and Microsoft Edge, with built-in device emulation for over 140 mobile and tablet devices. The server automatically handles browser binary installation on first use, eliminating manual setup. Configuration is highly flexible, supporting headless and headed modes, persistent user profiles for maintaining login sessions, proxy servers, viewport customization, and storage state management. With 28,000+ GitHub stars and deep integration into GitHub Copilot's coding agent, the Playwright MCP Server has become the de facto standard for connecting AI systems to the web browser, enabling use cases from automated testing and web scraping to form filling and end-to-end workflow verification.
28,522 starsGitHub
The GitHub MCP Server is the official Model Context Protocol server maintained by GitHub, giving AI assistants and coding agents direct access to the full GitHub platform. Originally developed by Anthropic as part of the MCP reference servers collection, it was adopted and expanded by GitHub into a standalone project at github/github-mcp-server. The server exposes over 50 tools spanning repository management, issue tracking, pull request workflows, code search, notifications, code security scanning, and GitHub Actions monitoring. AI agents can create and manage repositories, browse file contents, push commits, open and merge pull requests, conduct code reviews with inline comments, search code across all of GitHub, triage issues, manage notifications, and analyze security alerts from code scanning and secret scanning. The server supports configurable toolsets via the GITHUB_TOOLSETS environment variable, allowing you to enable only the capabilities you need and reduce context window usage. It also offers a read-only mode, lockdown mode for safely working with public repositories to guard against prompt injection, and dynamic toolset discovery. Authentication uses a GitHub Personal Access Token. The server can run locally via Docker using the ghcr.io/github/github-mcp-server image, as a remote HTTP server at https://api.githubcopilot.com/mcp/ with OAuth support, or via the legacy npm package for quick local testing. It integrates natively with Claude Desktop, Claude Code, VS Code, Cursor, and JetBrains IDEs.
27,690 starsFirecrawl
The Firecrawl MCP Server is the official Model Context Protocol integration from Firecrawl that brings production-grade web scraping, crawling, and structured data extraction directly into AI assistants like Claude, Cursor, Windsurf, and VS Code. Firecrawl is backed by Y Combinator and trusted by over 80,000 companies, with its core open-source project earning more than 88,000 GitHub stars. The MCP server itself has accumulated roughly 5,700 stars and receives over 8,800 weekly npm downloads, making it one of the most widely adopted web scraping servers in the MCP ecosystem. At its core, the server converts any publicly accessible website into clean, LLM-ready markdown or structured JSON by stripping ads, navigation elements, footers, cookie banners, and other boilerplate content. It handles JavaScript-rendered single-page applications, dynamically loaded content, and even PDF and DOCX documents without requiring the developer to manage headless browsers or proxy infrastructure. Firecrawl covers approximately 96 percent of the web without proxies and maintains a 95.3 percent scraping success rate with an average response time of seven seconds. The server exposes twelve MCP tools organized around five core capabilities. The scrape tool extracts content from individual pages with support for markdown, JSON, and screenshot output formats. The batch_scrape tool processes multiple URLs in parallel for high-throughput extraction workflows. The crawl tool traverses entire domains with configurable depth limits, URL filtering, and deduplication. The map tool discovers all indexed URLs on a site without requiring a sitemap. The search tool performs web searches with geographic targeting and time-based filtering, optionally scraping the full content of each result. The extract tool uses either cloud AI or self-hosted LLMs to pull structured data matching a developer-defined JSON schema. The agent tool conducts autonomous multi-step research by browsing multiple sources and synthesizing findings. Four browser tools (create, execute, delete, list) provide persistent Chrome DevTools Protocol sessions for interactive automation tasks like form filling, clicking, and scrolling. Configuration requires a single FIRECRAWL_API_KEY environment variable for cloud usage, with an optional FIRECRAWL_API_URL for self-hosted deployments. The server includes built-in resilience features: automatic retry with exponential backoff (configurable up to three attempts with delays from one to ten seconds), rate limit handling, and credit usage monitoring with configurable warning and critical thresholds. Installation takes a single command via npx, and the server supports both STDIO and Server-Sent Events transports for compatibility with remote and local MCP client configurations. A free tier provides 500 scraped pages to get started, with paid plans scaling from hobby to enterprise usage levels.
5,760 starsExa MCP Server
Exa is a search engine built specifically for AI — not humans. Unlike Google, which ranks pages by popularity and backlinks, Exa uses neural/semantic search to understand the meaning behind a query and return the most relevant content for machine consumption. Its API returns clean, structured text rather than HTML-heavy pages, making it ideal for feeding context into large language models. The Exa MCP Server exposes Exa's full search platform directly inside any MCP-compatible AI assistant or agent framework. Once installed, agents gain access to real-time web search, code context retrieval, company intelligence, deep research workflows, full-page crawling, and people/professional profile discovery — all through natural language tool calls. What sets Exa apart for AI developers is its combination of speed (sub-180ms latency), accuracy (outperforms competitors on FRAMES and Tip-of-Tongue benchmarks), and token efficiency (highlights reduce LLM costs by over 50%). The server ships with three tools enabled out of the box — general web search, code context from GitHub and Stack Overflow, and company research — while advanced tools like deep researcher agents, domain-filtered search, and people search can be unlocked as needed. Exa is SOC 2 Type II certified and supports zero data retention for privacy-sensitive workloads. A generous free plan is available, with API keys obtainable from the Exa dashboard.
4,013 starsTavily MCP Server
Tavily MCP Server gives AI agents the ability to search the live web, extract structured content from pages, map entire website architectures, and crawl sites systematically — all through the Model Context Protocol. Rather than relying on static training data, agents connected to Tavily can fetch real-time search results with advanced filtering, pull clean text and metadata from any URL, discover every page on a domain via site mapping, and perform deep multi-page crawls to build comprehensive knowledge bases on the fly. The server exposes four core tools. tavily-search delivers real-time web search with configurable depth, domain filtering, and optional image results. tavily-extract pulls structured data from individual web pages, stripping away boilerplate to return clean, usable content. tavily-map creates a complete structural map of a website, listing all discoverable URLs and their relationships. tavily-crawl systematically explores a site page by page, combining mapping and extraction into a single workflow for thorough data gathering. Deployment is flexible: run locally via npx with a simple API key, connect to the hosted remote server at mcp.tavily.com without installing anything, or use Docker for containerized environments. Authentication supports API keys (via query parameter or Authorization header) and a full OAuth flow for secure, token-based access. Default parameters like search depth, max results, and image inclusion can be set globally through environment variables or HTTP headers, ensuring consistent behavior across all requests. The server is compatible with Claude Desktop, Claude Code, Cursor, and any MCP-capable client, making it one of the most versatile web research tools available for AI agent workflows.
1,377 starsStripe Agent Toolkit
Ask Claude to create a payment link, email the invoice, and apply a 20% coupon -- in one prompt. The Stripe Agent Toolkit MCP server turns your AI assistant into a billing operations co-pilot that can read, write, and manage your entire Stripe account without you ever opening the dashboard. The server exposes 30+ tools covering the full billing lifecycle. Create and list customers, build product catalogs with tiered pricing, generate payment links for instant checkout, draft and finalize invoices with line items, manage subscriptions (create, update, cancel), process refunds, handle disputes, and create coupon codes. Utility tools like search_stripe_resources and fetch_stripe_resources let Claude query any Stripe object by ID or search term, and search_stripe_documentation gives it access to Stripe's knowledge base for answering integration questions on the fly. Stripe hosts a remote MCP server at mcp.stripe.com with OAuth authentication, so you can connect from Claude Desktop, Cursor, or VS Code without managing API keys locally. For local development or CI pipelines, run `npx -y @stripe/mcp --api-key=sk_test_...` and you are up in under a minute. Security is handled through Restricted API Keys -- you decide exactly which Stripe resources your AI assistant can touch, down to individual API methods. The real power shows up when you combine Stripe MCP with other servers. Pair it with GitHub MCP to build an end-to-end workflow: customer reports a billing bug in an issue, Claude reads the issue, queries Stripe for the customer's invoice history, identifies the problem, issues a refund, and posts the resolution back to the issue. Or connect it with a database MCP server to reconcile your internal records against Stripe's data. Beyond MCP, the toolkit also works as a standalone SDK for OpenAI Agents, LangChain, CrewAI, and Vercel AI SDK -- but the MCP server is the zero-config path for Claude users. This is an official Stripe product built on the Stripe Node SDK, backed by Stripe's security team, and MIT-licensed. If you run any kind of SaaS, e-commerce, or subscription business and use an AI coding assistant, this server pays for itself the first time you resolve a billing support ticket without leaving your editor.
1,359 starsBrave Search
The Brave Search MCP Server is the official Model Context Protocol integration from Brave Software that connects AI assistants to Brave's independent web search index. Unlike most search APIs that rely on Bing or Google indexes, Brave operates its own large-scale web index built with contributions from its privacy-preserving Web Discovery Project. The server exposes six specialized MCP tools: brave_web_search for comprehensive web searches with advanced filtering and custom re-ranking via Goggles, brave_local_search for discovering nearby businesses with ratings, hours, and contact details, brave_image_search and brave_video_search for rich media discovery with safety filtering and metadata, brave_news_search for current news articles with freshness controls, and brave_summarizer for AI-generated summaries from search results using Brave's summarization API. The server supports both STDIO and HTTP transport modes, making it compatible with all major MCP clients including Claude Desktop, Claude Code, Cursor, Windsurf, and VS Code. Configuration is straightforward with a single BRAVE_API_KEY environment variable, and the free tier provides 2,000 queries per month at no cost. Version 2.x of the server returns response objects that closely mirror the original Brave Search API structure, providing rich metadata including spell check suggestions, content safety classifications, and location-aware results. The package replaces the now-deprecated @modelcontextprotocol/server-brave-search and is actively maintained by Brave Software under the MIT license.
774 starsReddit MCP Server
A zero-configuration Model Context Protocol server that gives AI assistants full access to Reddit content without requiring API keys or authentication. Built on the redd scraping library, this server lets Claude, ChatGPT, and other MCP-compatible clients browse subreddits, search posts, read comments, and analyze user activity — all through structured tool calls. The server exposes tools for fetching front page posts, browsing specific subreddits by category (hot, new, top, rising), reading full post content with comments, searching across Reddit, and analyzing user profiles. All data is returned in clean, structured formats optimized for LLM consumption. What sets this server apart from other Reddit integrations is the zero-config approach. Most Reddit MCP servers require OAuth credentials, API keys, or a registered Reddit app. This one uses web scraping via the redd library, which means you install it, add it to your MCP config, and it works immediately. No Reddit developer account needed. The server supports both stdio and SSE transports, making it compatible with Claude Desktop, Claude Code, VS Code with MCP extensions, and other MCP clients. It runs as a standard Node.js process and can be launched directly via npx for instant setup. Practical use cases include market research by monitoring subreddit sentiment, content research by finding discussions on specific topics, competitive analysis by tracking mentions of products or brands, and community engagement analysis by understanding what content resonates in specific communities.
85 starsUpstash MCP Server
Upstash's official MCP server lets AI assistants manage your Redis databases through natural language. Instead of switching between your AI coding environment and the Upstash Console, you tell Claude or ChatGPT to create a database, run commands, check stats, or configure backups — and the MCP server translates that into Upstash Developer API calls. The server exposes a focused set of Redis operations: create and delete databases, execute single or batch Redis commands, pull usage statistics, manage daily backups, update regions, and reset credentials. It connects via stdio transport for local MCP clients or HTTP transport for web-based integrations. Authentication uses your Upstash email and API key from the Upstash Console. Once configured, your AI assistant has full programmatic access to your Redis infrastructure without you touching a dashboard. The practical value is clearest in development workflows. You're building a feature, realize you need a new Redis cache, and instead of context-switching to a browser, you say 'create a Redis database in us-east-1' and keep coding. Need to debug a cache issue? Ask the assistant to run GET/SET commands directly. Want to check if your rate limiter is working? Pull stats without leaving your editor. Currently limited to Redis operations — no QStash, Vector, or Kafka support yet. But for teams already using Upstash Redis, this removes a significant amount of dashboard friction from daily development.
52 starsZapier MCP Server
Zapier MCP Server is a remote Model Context Protocol server that bridges AI assistants with more than 8,000 applications and over 40,000 discrete actions through a single, standardized interface. Built and maintained by Zapier, one of the most widely adopted automation platforms in the world, this MCP server transforms conversational AI tools into operational powerhouses capable of executing real workflows across business-critical services. Rather than requiring developers to build and maintain individual API integrations for every third-party service, Zapier MCP exposes each configured action as a dedicated callable tool that any MCP-compatible AI client can invoke using natural language. For example, enabling the Gmail Send Email action gives your AI assistant a gmail_send_email tool that accepts parameters like recipient, subject, and body, letting the AI compose and dispatch messages on your behalf. The same pattern applies across the entire Zapier ecosystem, from sending Slack messages and creating Asana tasks to updating HubSpot deals and adding rows to Google Sheets. Authentication is handled through two pathways: API keys for personal use and local development, and OAuth for production applications where end users connect their own Zapier accounts. Every server instance includes built-in meta-tools that let the AI discover available actions and understand what each one does before invoking it. Enterprise-grade security features such as encrypted connections, rate limiting, and activity audit logs ensure that sensitive workflows remain protected. Zapier MCP is compatible with leading AI clients including Claude Desktop, ChatGPT, Cursor, Windsurf, and any other client that supports the streamable HTTP transport. Configuration is managed through a web dashboard at mcp.zapier.com, where users browse available apps, enable specific actions, and generate connection credentials. For local AI clients like Claude Desktop, the mcp-remote npm package serves as a bridge between the remote server and the local MCP client protocol. The server operates on a task-based billing model integrated with standard Zapier plans. Each MCP tool call consumes two tasks from your plan quota, and the free tier provides 100 tasks per month, allowing up to 50 MCP invocations. Paid plans scale significantly higher, making this server suitable for both individual experimentation and team-level production automation.
19 stars