MCPForge Alternatives

MCPfinder is a free, open-source discovery and installation layer designed for AI agents to locate and configure Model Context Protocol (MCP) servers. It serves as an MCP server itself, enabling AI assistants to programmatically search multiple MCP registries, inspect trust signals, and generate install-ready JSON configurations for downstream MCP servers. The tool is intended primarily for AI agents, with humans only needing to install it once; after installation, the AI assistant can use MCPfinder repeatedly to expand its capabilities on demand. Key features include multi-registry search, aggregating MCP servers from the Official MCP Registry, Glama, and Smithery into a unified search interface. MCPfinder allows AI agents to find candidate servers by keyword, technology, or use case, inspect trust signals and required environment variables, and generate client-specific configuration snippets suitable for platforms such as Claude Desktop, Cursor, Claude Code, Cline, or Windsurf. The tool provides structured outputs, including confidence scores, recommendation reasons, warning flags, and install complexity, helping AI assistants make informed decisions about which servers to recommend or install. MCPfinder is delivered as an MCP server that can be added to any MCP-compatible AI client. Installation is flexible, supporting direct execution via NPX (no install needed), global installation with NPM, or project-based integration. Once installed, it operates as an always-up-to-date agent-facing layer, bootstrapping from published snapshots and syncing from upstream registries. This ensures that AI agents have access to the latest available MCP servers and their metadata, with freshness signals provided by snapshot manifests. 0 and is free to use, with the source code available for inspection, contribution, or forking. MCPfinder is built by the community for the community, and as of its latest published snapshot, it aggregates tens of thousands of MCP servers across multiple registries. Its focus on agent-native discovery, trust signal inspection, and install-ready configuration distinguishes it as an AI-centric solution within the ecosystem of MCP tools.

MCP Servers is a marketplace and discovery platform that enables users to search for and explore a variety of MCP servers, clients, command-line tools, integrations, and reusable workflows. The platform is organized into categories such as Developer Tools, AI & Agents, Cloud & Infrastructure, Memory & Knowledge, and Media & Design, making it easier to find production-ready servers and related resources. The service highlights featured servers and clients, including those supporting payments, analytics, code editing, and database management. It lists both remote and local servers, as well as official and agent-ready CLI tools for tasks like JavaScript runtime management, database querying, audio/video processing, and cloud development. MCP Servers also showcases reusable prompts and workflows, referred to as "loops," which can automate tasks such as testing, accessibility auditing, code coverage, and continuous integration monitoring. In addition to servers and CLI tools, the platform features agent skills tailored for AI agents, covering areas like brainstorming, systematic debugging, writing plans, and skill invocation. These resources are designed to facilitate the connection of AI applications to various tools, data sources, and automated workflows. The platform also provides a feed of trending and newly published servers, offering insights into what the community is installing and using. MCP Servers is aimed at developers and those building AI-powered applications who need to integrate diverse tools and automate workflows. The platform is web-based and includes a registry of available servers, clients, and integrations.

PowForge MCP is middleware designed for MCP server operators who need to control access to costly tools by combining payment gating and reputation checks. It addresses the challenge of preventing abuse by unknown or untrusted agents, ensuring that only callers who have both paid a Lightning invoice and meet a minimum reputation threshold can access protected endpoints. The middleware integrates a Lightning Network L402 payment flow with a Schnorr-signed depth-of-identity (DoI) score, allowing operators to reject callers below a configurable reputation threshold and to charge per tool call. The platform provides an Express middleware package, installable via npm, that can be dropped in front of any MCP tool route. When a caller accesses a gated route, the middleware issues a 402 response with a Lightning invoice. Upon payment, the caller replays the request with the required authorization fields and their public key. The middleware then queries an external oracle for the caller’s DoI score, which is cached for five minutes per public key. Only callers whose score meets or exceeds the specified threshold are allowed to proceed; others receive a 403 response. The system uses one-time-use macaroons for authentication, which expire after a configurable period. For non-Express environments, PowForge MCP offers a tool factory that returns a plain MCP tool definition with the gate logic embedded. This allows integration with any MCP server, including those using the official MCP SDK or custom HTTP transports. The middleware is suitable for operators managing tools with significant operational costs, such as image rendering, web scraping, or code execution, and for public MCP tool authors who want to deter abuse from throwaway identities while allowing genuine users to pay once and continue using the service. Reputation thresholds are configurable, with suggested minimum scores corresponding to different levels of trust and operational risk. The middleware can be set to fail closed if the reputation oracle is unreachable or to allow fallback behavior. Installation is via two npm packages, and a working server example is available for cloning. The approach is intended to provide both payment enforcement and sybil resistance for open or high-cost MCP tools.

alpha-forge-mcp is an open-source MCP server that exposes the AlphaForge CLI to AI coding tools like Claude Code and Cursor. It facilitates quant trading, backtesting, and integration via the Model Context Protocol for developers and AI tool builders.

MCP Score is an independent trust registry for MCP servers. It assigns MCPScore values on a 0–100 scale and is described as security-first, with the stated goal of showing when things get worse. The registry scores each server on security, freshness, adoption, quality, and trust. Its methodology is public and versioned, and the site shows how those factors combine into the overall score. The homepage also surfaces several views built around those scores, including top by score, most downloaded, new arrivals, security risks, and score drops. It includes a compare function, search, and filters for registry source and score thresholds. The directory covers 46,736 scored servers and says it is updated nightly. Listings shown on the page include the server name, registry source such as npm or PyPI, weekly download counts, and star counts where available. The browse area groups servers by categories such as Databases, Dev Tools, Web Scraping, Automation, AI & ML, Finance, Monitoring, Communication, Search, Storage, and Other. The site describes itself as independent and states that it takes no vendor money. It is delivered as a website at timeahead.in, with paths for servers, blog, and about.

mcp-forger is an open-source tool that enables developers to convert any application into an MCP server with the help of AI. It offers both a CLI and a desktop plugin, integrating with Claude for enhanced automation and code generation.

MCP Pre-Flight is a submission audit tool for MCP servers being prepared for Claude and OpenAI directories. It is meant to catch the kinds of mismatches that reviewers reject before a submission goes in, and it presents that check as a three-layer process covering wire-level checks, per-tool invocation, and source-side AST analysis. Its wire-level tests cover OAuth 2.1 with PKCE, dynamic client registration, authorize, token exchange, refresh, and MCP protocol endpoints including initialize, tools/list, and tools/call, along with malformed-body and unknown-method resilience. At the tool level, every safe-to-invoke tool is actually called with synthesized arguments, and idempotent tools are probed by calling them twice and comparing the results. The source-side analysis is optional and accepts a GitHub repo URL; it runs an AST check to compare handlers against annotations, checks that the README mentions every declared tool, and analyzes response language for vocabulary that contradicts annotations. Findings are tagged by directory, with each issue marked for Claude or OpenAI. The service is aimed at developers submitting MCP servers to those directories. Its FAQ says the input should be the HTTPS endpoint of a deployed MCP server using Streamable HTTP transport, and that the server should respond to JSON-RPC tools/list. It also notes that if OAuth is present, the full auth flow is tested, and that an optional test bearer token unlocks authenticated MCP checks. Source probing uses the GitHub API, and the page says pre-flight only probes the live endpoint; if source is checked later through the M8ven Trust Index, the repository is cloned in an isolated sandbox and then deleted. MCP Pre-Flight is offered as a free web tool. The page also states that Claude and OpenAI directories require OAuth for listed MCPs, and it mentions an OpenAI domain challenge file at /.well-known/openai-apps-challenge for OpenAI submissions.

CLI scaffold tool for mcp-forge — generate MCP server projects in seconds

mcp-mcts is an open-source CLI tool for local MCP security scanning, attack chain analysis, inventory management, and CI integration. It supports optional Semgrep and LLM analysis, making it suitable for security engineers and DevOps teams focused on codebase security.

mcpvet is an open-source CLI tool that combines testing, linting, and contract validation for MCP servers. It integrates with pytest and ESLint, streamlining the development and quality assurance process for developers working with the Model Context Protocol.

MCP Playground is an interactive playground for the Model Context Protocol. It is built to test remote MCP servers in the browser, letting users paste a server URL, inspect what it exposes, and run tools live without installation. The service supports remote endpoints over Streamable HTTP, SSE, and WebSocket, and it says it can auto-detect the right protocol or connect directly via ws://. It presents the tools, resources, and prompts a server exposes, shows full JSON Schema, and runs them through auto-generated forms. The page also lists a schema linter for grading MCP servers A–F with 15+ lint rules, and a quality dashboard for browsing registry-wide grades. A public REST API is mentioned for inspecting, linting, and health-checking MCP servers, and the site also offers an in-browser sandbox for stdio-based npm MCP servers via WebContainers. MCP Playground is aimed at people working with MCP servers, including server authors and developers. The page also offers a public registry view with 852+ servers, examples for testing live servers, and an embeddable “Try in MCP Playground” badge for README files. It says authenticated servers can be used with a user’s own API key and authorization headers. The site is delivered in the browser and requires no installation. It is built on the official MCP Registry and the official MCP SDK, and it is open source under the AGPL v3 license. The page also links to an associated mcpx CLI and CI workflow for running the linter in terminals and pipelines, with mentions of GitHub Actions, GitLab CI, and other CI systems.

MCP Playground is a free browser-based testing tool for the Model Context Protocol. It is meant for testing MCP servers, clients, and tools directly in the browser, with no installation or configuration required. The service is aimed at developers and is described as a way to validate protocol implementations, test tools before deployment, and explore MCP servers and resources. The site supports connecting to remote MCP servers from a browser and browsing their tools, prompts, and resources. It offers real-time request and response logs, full request/response JSON for each tool call, per-call latency, and credit cost details. The page also mentions an MCP Security Scanner, an MCP Token Counter, a Config Generator, MCP Tutorials, and MCP Recipes. In addition, the MCP Agent Studio lets users build, test, and ship MCP-powered AI agents in the browser, compare up to four AI models in parallel, and export an agent as a production API with bearer authentication and a server-side locked prompt. Transport support includes HTTP, SSE, Streamable HTTP, and local stdio/CLI processes. The site says users can connect up to four servers at once in one chat, with tools merged automatically, and that agents can be saved and reused as named configurations. It also lists browsing 10,000+ MCP servers, including Supabase, Playwright, Figma, and GitHub, and says users can add any server with one click. For local tools, it mentions running npx or uvx alongside remote HTTP servers and injecting environment variables for authentication in the UI. MCP Playground is described as free, with no sign-up or credit card needed, and the Agent Studio mentions free credits on sign-up. The page also says features remain free and no registration is required. It presents itself as a browser-based MCP testing tool rather than a desktop app or local install.

Security scanner for MCP (Model Context Protocol) servers - pentest your AI agent's tool connections

MCPSafe is a free MCP server security scanner for checking whether a server is safe to install. It is built for developers vetting MCP servers before installation and for registry operators publishing safe catalogs. The service describes its checks as being backed by static analysis and five LLMs, and it offers both a fast verdict in about three minutes and a deeper LLM-judge consensus run in about twenty minutes. Its checks cover several named areas. Typosquatting flags lookalike package names, unverified publishers, and unpinned dependencies before they reach an install chain. Static analysis is used to detect command injection, SQL injection, SSRF, path traversal, and hardcoded secrets. The LLM consensus layer uses five independent LLM judges to uncover tool poisoning, silent rug pulls, indirect prompt injection, and obfuscated intent that pattern matching misses. A permission audit examines each tool’s real-world reach and flags excessive permissions, weak authentication, and network exposure beyond its stated purpose. MCPSafe accepts multiple source types. The scan flow says users can paste an MCP source from GitHub, an npm scoped package, or a pip package, and the FAQ adds support for GitHub URLs, npm packages, PyPI packages, Docker images from Docker Hub or GHCR, and MCP registry IDs. It normalizes those inputs, runs typosquat, static, behavioral, readiness, and 5-LLM consensus analysis in parallel, and returns an AIVSS 0–10 score with per-tool findings, CWE mapping, and copy-safe config. It also notes that already-scanned servers return in under a second and that it re-verifies on every new commit. The scanner is free to use, with no credit card required and no signup required for scanning. Signed-in users get higher rate limits and scan history. It is delivered as a web service, includes a live SVG badge for scanned packages, and is marked as made in Germany. The page also describes public packages as free and private repos as available for teams shipping their own.

mcp-check-security is an open-source command-line tool that performs offline security checks on Model Context Protocol (MCP) server configurations. It helps developers and security engineers identify potential vulnerabilities and misconfigurations before deployment. The tool is designed for use in AI infrastructure environments.

mcp-grade is a command-line tool that acts as a Lighthouse or ESLint for MCP servers. It grades any MCP server with an A–F score, probes it live for capabilities, and maintains a public leaderboard ranking the ecosystem. It is designed for developers building or evaluating Model Context Protocol implementations.

MCP Scoreboard is a web-based analytics platform that provides independent quality scores for public MCP servers. Users can browse, compare, and track server reliability, access data via API or CLI, and receive score alerts. It is designed for MCP server operators and ecosystem participants.

MCP Meta-Orchestrator for Forge — routes, secures, budgets, and caches tool calls across N upstream MCP servers.

MCP Servers is a WordPress plugin designed to assist site administrators in configuring and exposing MCP servers with selected abilities on their WordPress sites. The tool facilitates integration between WordPress and AI agents by enabling secure authentication and permission management for server access. The plugin supports the configuration of user roles and authentication methods for AI agents interacting with the MCP server. It provides guidance on creating dedicated user accounts, such as a user named "mcp" with an appropriate role, and on generating application passwords for Basic Auth. For AI agents requiring OAuth2 authentication, the documentation suggests installing an additional plugin, such as WP Oauth Server, to enable OAuth2 support within WordPress. The plugin also addresses compatibility considerations for specific AI agents, including Mistral and Claude Desktop, and offers instructions for connecting via local proxies and token-based authentication when necessary. Installation of MCP Servers is handled through the standard WordPress plugin upload process, and the plugin is set to update automatically after the initial installation. Logging and debugging features are available: if WordPress debugging is enabled, internal events are recorded in the error log, and when logging is activated, MCP server activities are accessible via a dedicated logs page within the plugin. MCP Servers is available for download from GitHub.

Prufa is an AI QA engineer for web products that checks signup, login, and checkout flows after deployment. It is built to catch breaks in minutes after every deploy, using a real browser and machine verification rather than waiting for customers to report a problem. Its reports separate verified findings from advisory findings. Verified findings cover deterministic checks such as redirects, HTTP status codes, and beacon events, and the page shows examples including a payment redirect returning 404, a GA4 conversion beacon not firing, and a login flow passing all four steps. Advisory findings are marked as opinions, such as a form field that appears to lack a visible label or a consent banner that may overlap a primary call to action. Prufa also includes Gremlin mode, which runs money flows under adversarial conditions by mutating inputs, timing, and edge-case paths; the page describes this as chaos QA and says each run produces a most valuable finding. The service runs a real-browser audit after a URL is pasted, and the report appears in the browser in about 60 seconds. It says no account, no npm install, and no Playwright config are needed for the first audit. After a site is connected, every push to the main branch triggers a re-run of the checks via a deploy webhook, and a Slack alert can fire with a screenshot and step trace when a previously passing flow fails. The page also shows a report view with step-by-step runs and named assertions. A free audit is offered with no credit card required and no signup required to run it. The product is presented as an AI QA engineer for web products.

BoltMCP is a platform designed for organizations seeking to create and manage secured MCP (Modular Command and Control Platform) servers entirely on-premises. The tool addresses the need for secure, customizable integration of AI agents with organizational data and APIs, without relying on external gateways or cloud services. Its architecture allows teams to build token-efficient MCP servers that fit specific use cases, leveraging existing infrastructure and workflows. The platform emphasizes seamless integration with an organization’s current systems. It supports the use of existing identity providers for authentication, enables application of custom RBAC (role-based access control) policies for both users and tools, and connects with in-house APIs for data retrieval and actions. BoltMCP also integrates with established observability platforms for telemetry, ensuring that monitoring and oversight remain within the organization’s control. A key feature of BoltMCP is its LLM-friendly design, abstracting complex API workflows into simplified, token-efficient tools and resources that large language models can easily interpret. The system employs progressive disclosure by default, providing AI agents with only the necessary information at the right time to avoid overloading context windows and reducing unnecessary token usage. The platform claims measurable improvements in efficiency and recall, such as saving 25,000 tokens on average per task compared to pre-loading resources, and achieving 35% better recall of task-specific tools over naive retrieval-augmented generation (RAG) methods. Additional capabilities include built-in version control, enabling teams to manage and track changes to their MCP servers, and a kill switch feature that allows instant disabling of servers, revoking user access, or restricting specific tools as needed. BoltMCP supports deployment on the user’s own infrastructure, offering flexibility and avoiding vendor lock-in or cloud dependencies. The platform is positioned for organizations that require secure, efficient, and customizable AI integrations within their existing IT environments.

mcpfuzz is an open-source command-line tool for dynamically scanning Model Context Protocol (MCP) servers. It actively probes live servers with exploit payloads to identify vulnerabilities, supporting security researchers and penetration testers in automating MCP server assessments.

MCP Gateway is a no-code tool for turning OpenAPI-based APIs and selected PostgreSQL or SQL Server database scopes into MCP tools for AI agents. It is built for publishing an MCP surface from existing sources, with editable descriptions and a testing workflow in Gateway Chat. The product imports a Swagger/OpenAPI specification or connects to a database scope, then generates MCP tools from that source. It supports reviewing the generated tools, choosing which operations agents can see, hiding others, and rewriting descriptions so tool choice is clearer. The Gateway also lets users test the result in Gateway Chat, where it selects published MCP tools and shows the call. Its published surfaces can expose APIs as MCP tools, and for databases it can publish PostgreSQL or SQL Server scopes as metadata-only schema tools plus validated QueryPlan or SQL executors. Several controls remain on the Gateway side rather than in prompts or local client configuration. The page says the Gateway keeps authentication, validation, limits, and logs server-side, and that credentials stay server-side. It also mentions redacted logs, along with usage views that show call volume, success rate, latency, and tool analytics. The product flow shown on the page includes importing the source, generating tools, tuning descriptions, testing in Gateway Chat, and then connecting external MCP-compatible clients such as Claude, Cursor, VS Code, or any MCP-compatible client. Pricing is listed in three plans. Free is described as for trying the gateway and small projects, with $0 forever, up to 1 integration, 1,000 tool calls per month, and 30 tool calls per minute. Pro is $9 per month or $69 per year, with up to 10 integrations, 50,000 tool calls per month, and 120 tool calls per minute. Team is $199 per month or $499 per year, with up to 50 integrations, 200,000 tool calls per month, and 300 tool calls per minute.

RMCP is a searchable directory of public Model Context Protocol servers. It aggregates thousands of remote MCP endpoints so that both human users and AI agents can locate suitable servers and tools for specific tasks without needing a registry API key. The directory contains 3,072 servers, 3,129 endpoints, and 58,366 known tools. It supports fuzzy search and tool-aware queries across categories that include AI and machine learning, analytics, developer tools, e-commerce, content tools, data collection, and dozens of others. Visitors can browse recently added entries, view details such as the number of tools per endpoint, and inspect individual servers such as those providing blog-post access, job data, or curated web apps. It functions as an instantly updated remote MCP directory. Searches can target popular areas such as developer tools, e-commerce, analytics, and AI search. The service is delivered through a web interface at rmcp.dev that presents the catalog, category lists, and direct links to explore each entry.