The Best PulseMCP Alternative for Verified Agent Skills in 2026

PulseMCP is a large public directory of Model Context Protocol servers. Its catalog pages show 3,856+ servers browsable with filters for remote availability, classification (official providers vs community), Anthropic references, last-updated date and popularity, and the site markets itself as 21,690+ servers updated daily. Each listing carries structured metadata: classification, estimated visitor or download counts, release dates, popularity rankings on a global leaderboard, and in many cases a standardized server.json file with installation instructions and configuration options. PulseMCP also runs an editorial layer — trending server lists (Figma, Notion, GitHub, Supabase and dozens more), guest posts from MCP founders on topics like pricing a paid server, and category pages such as RAG. Its strength is breadth and discoverability; its limitation is that it classifies and ranks servers but does not security-scan, license-check or benchmark them, and new submissions are currently paused while the ingestion pipeline is reworked.

pricing See pricing

• + One of the largest MCP server catalogs available, with thousands of servers browsable and marketed as 21,690+ updated daily
• + Structured per-server metadata: classification, release dates, popularity rankings, estimated usage and server.json configs
• + Strong discovery UX: trending lists, category filters, remote-available and official-provider filters
• + Free to browse, with many listed servers (such as Planning Center's) entirely free to use
• + Active editorial coverage of the MCP ecosystem, including founder case studies

• − No security scanning, license checking or quality benchmarking of listed servers or skills
• − New server submissions and listing changes are currently paused while ingestion is reworked
• − Popularity and visitor figures are algorithmic estimates, not verified usage data
• − No skill-level verification — nothing tells you whether a given skill actually helps or is risky

https://www.pulsemcp.com

SkillzBench is AIQ Labs' free, independent directory of AI agent skills plus a free MCP tool that connects any agent to the skills you choose. Agent skills — SKILL.md instruction packages for Claude Code, Codex, Cursor, Gemini CLI and chat-app agents — have proliferated with no quality gate: published research found roughly 26% carry at least one vulnerability. SkillzBench is built around five pillars: Discover (continuously crawling and deduping skills from GitHub), Verify (pinning every version to a commit SHA, validating, license-checking, security-scanning and quality-judging), Benchmark (testing every skill in a sandbox with the same model, with and without the skill, and publishing the Lift in percentage points), Equip (one free MCP server plus per-agent loadouts where you toggle skills and tools per agent), and Stay Current (tracking upstream changes, re-scanning, re-benchmarking and notifying you). A 0-100 trust score combines security, quality, license, maintenance and community signals; Lift is always shown separately, and money never touches scores. The product is pre-launch — the waitlist and newsletter are live while the platform is being built.

pricing 0

• + Everything is free — directory, MCP connector, benchmarks and alerts — with ad revenue that never influences scores or listings
• + Addresses a real, quantified risk: published research found ~26% of agent skills carry at least one vulnerability
• + Verification is concrete: commit-SHA pinning, security scans, license checks and quality judgments per version
• + Lift benchmarks are measured in a controlled sandbox and always displayed separately from ratings
• + Per-agent loadouts give precise control over what each agent can see and use

• − Pre-launch: the product is not yet publicly available; the waitlist and newsletter are live
• − Focused on agent skills rather than the full MCP server universe, so catalog breadth differs from a general server directory
• − No published live skill counts or benchmark results yet

https://skillzbench.com

MCP Hive is a commercial marketplace-style subscription that bundles access to participating MCP servers behind a single plan. Instead of evaluating and subscribing to servers individually, users pay one monthly fee for a call allowance across all participating MCPs, with some priced servers charging their own per-call costs covered by monthly credits. The Starter plan costs $19.99/month for 8,000 MCP calls (roughly $0.0025 per call), Professional is $39.99/month for 16,000 calls plus a $20/month credit for priced MCPs, and Enterprise starts at $249.99/month with pay-as-you-go beyond credits at $20 per additional 8,000 calls under a custom spending cap. All plans include a 30-day free trial with no credit card required, and a console tracks calls used and remaining per billing period. On the provider side, MCP Hive retains 30% of what a server earns and pays out 70%. It is an access-and-billing aggregator rather than a verification or benchmarking service.

pricing 19.99/month · free trial

• + 30-day free trial on all plans with no credit card required
• + Predictable per-call economics — Starter works out to roughly $0.0025 per call
• + Enterprise tier scales cleanly with pay-as-you-go increments and a spending cap you set
• + No setup fee and no separate per-call fee for participating servers included in the plan

• − Subscription starts at $19.99/month versus free alternatives for discovery and basic access
• − Priced MCP servers can charge beyond the included credits
• − No published security scanning, license checks or skill benchmarking
• − Access limited to participating servers on the platform

https://mcp-hive.com

Ref is a premium documentation-search MCP server built specifically for AI coding agents, and one of the first standalone paid MCP products. It exposes two tools — ref_search_documentation and ref_read_url, each consuming one credit — letting coding agents search a curated public index of library and package documentation plus private indexes of team PDFs and GitHub repos. Its differentiation is search precision optimized to avoid wasting the agent's context window, plus hybrid public-private indexing for team documentation. Pricing uses 200 free non-expiring credits, then $9/month for 1,000 credits (about $0.009 per search), charging only for searches rather than indexing. Within three months of launching this model, Ref reported thousands of weekly users and hundreds of paying subscribers. Infrastructure-wise it runs on Turbopuffer search indices with an LLM retrieval pipeline and a web crawler that is its highest cost. Ref serves individual engineers, small teams sharing one API key, and agent developers.

pricing 9/month · free trial

• + 200 free credits that never expire before any payment
• + Precise documentation search purpose-built for coding agents
• + Supports private team indexes, not just public docs
• + Proven traction: thousands of weekly users and hundreds of subscribers within three months

• − Narrow scope — documentation search only, not a skills directory or verification layer
• − Costs scale with agent usage; heavy autonomous agents can burn credits fast
• − Everything else in the free ecosystem makes a paid model a harder fit for casual users

https://ref.tools

Context7 provides a free MCP server with basic RAG (retrieval-augmented generation) for documentation, drawing primarily from public GitHub repositories. In the competitive landscape mapped by Ref's founder, Context7 occupies the free end of the documentation-search spectrum: it gives coding agents a no-cost way to pull relevant library documentation into context, while paid options like Ref and the Tavily/Exa search APIs differentiate on retrieval precision, private indexing and infrastructure quality. For small businesses and individual developers experimenting with AI coding agents, Context7 is a zero-cost starting point that requires no subscription and no per-call budgeting. Its trade-offs mirror that positioning: it offers basic RAG rather than precision-tuned retrieval, focuses on public repository documentation rather than private team knowledge, and — like most free community offerings — does not publish security scans, license checks or benchmark results for what it serves.

pricing 0

• + Completely free to use
• + Zero-friction way to give coding agents documentation context
• + Widely recognized as the free baseline in the documentation-search category

• − Basic RAG rather than precision-tuned retrieval
• − Primarily public GitHub repo documentation — no private team indexes
• − No published security scanning or quality benchmarking

https://context7.com

Tavily offers agent-focused web search and scraping services, charging roughly $0.01 per search with volume discounts, and maintains a popular MCP server that lets AI agents query the live web directly. In the MCP ecosystem it serves as one of the standard paid benchmarks: Ref's founder explicitly used Tavily and Exa as the pricing and capability comparables when positioning a premium documentation-search product. Where Context7 covers static documentation via RAG, Tavily covers live web retrieval — useful for agents that need current information beyond any fixed documentation index. For small businesses building agents that research, monitor or gather web data, Tavily is a mature, widely integrated option. Its limitations for the verified-skills use case are structural: it is search infrastructure, not a directory or verification service, so it tells you nothing about whether a given agent skill is safe, licensed or effective, and its per-search costs need budgeting as agent usage scales.

pricing 1 tiers

• + Live web search and scraping purpose-built for agents
• + Established MCP server with wide ecosystem adoption
• + Volume discounts available for heavier usage

• − Per-search costs (~$0.01) add up for autonomous agents making thousands of calls
• − Search infrastructure only — no skill verification, licensing or benchmarking
• − Not a directory, so it doesn't help you choose between skills or servers

https://tavily.com