PulseMCP vs SkillzBench: Which Agent Skills Platform Should You Trust?
PulseMCP is a live, free-to-browse MCP server directory listing 21,690+ servers with remote-endpoint details, server.json files, and popularity estimates, though new submissions and listing changes are currently paused. SkillzBench is a free, independent directory of AI agent skills with a free MCP connector, and it is building version-pinned, security-scanned, sandbox-benchmarked skill reviews — but it is pre-launch, so nothing is publicly live yet. If you need a usable MCP server catalog today, PulseMCP is the only one of the two you can actually browse; if you want to know which agent skills genuinely help versus which are risky, SkillzBench is the one building that answer.
TL;DR
- PulseMCP is a live MCP server directory claiming 21,690+ servers, updated daily, with filters for remote availability, official providers, and community classifications.
- PulseMCP's directory pages are free to browse, and individual server profiles show remote endpoint URLs, auth types, transport protocols, and cost labels such as 'Free'.
- PulseMCP has paused new server submissions and listing changes while it reworks how it ingests and manages listings.
- 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 — currently pre-launch with a live waitlist.
- SkillzBench pins every skill version to a commit SHA, validates, license-checks, security-scans, and quality-judges each skill, responding to published research finding ~26% of skills carry at least one vulnerability.
- SkillzBench plans to benchmark every skill in a sandbox with the same model, with and without the skill, and publish the Lift in percentage points — always shown separately from ratings.
- Everything on SkillzBench will be free; revenue comes from advertising around blog articles, and ad money never influences scores or listings.
Head to head
| Dimension | PulseMCP | SkillzBench |
|---|---|---|
| Pricing | Directory pages free to browse; individual listed servers may carry their own costs shown on profiles | Everything free — directory, MCP connector, benchmarks, alerts; funded by blog advertising that never touches scores |
| Availability | Live today — 21,690+ servers browsable, though new submissions and listing changes are paused 🏆 | Pre-launch — waitlist/newsletter live, product being built |
| Catalog focus | MCP servers — tools and data connectors across the ecosystem (GitHub, Slack, Notion, RAG, databases, and more) | AI agent skills — SKILL.md instruction packages for Claude Code, Codex, Cursor, Gemini CLI and chat-app agents, crawled and deduped from GitHub |
| Verification & security | Listings, metadata, and standardized server.json files; no published security scanning of listed servers | Commit-SHA version pinning, validation, license checks, security scans, and quality judging for every skill 🏆 |
| Benchmarking | Popularity and visitor estimates per server; no performance benchmarking | Sandbox tests with the same model, with and without each skill, publishing Lift in percentage points 🏆 |
| Setup & integration | Copy/paste remote endpoint URLs into your MCP client; server.json files document configuration | One free MCP server connects any agent to chosen skills; per-agent loadouts limit what each agent sees |
| Freshness & maintenance | Directory updated daily, but submissions and listing changes currently paused during an ingestion rework | Planned continuous tracking of upstream changes with re-scan, re-benchmark, and notifications |
| Trust model | Provider classifications (official, community) and estimated popularity rankings | 0-100 trust score from security, quality, license, maintenance and community signals; money never touches scores 🏆 |
The contenders
PulseMCP is a directory of Model Context Protocol (MCP) servers that claims 21,690+ listings updated daily. It is built to work with a variety of MCP clients and offers filters for remote availability, classification (Anthropic References, Official Providers, Community), and sorting by last updated, popularity, and release date. Individual server profiles include a standardized server.json file with installation instructions, the exact remote endpoint URL to paste into an MCP client, auth type, transport protocol, and cost labels such as 'Free'. PulseMCP also estimates visitor or download volumes per server and popularity rankings, and it publishes ecosystem guest posts, such as a case study on pricing a standalone paid MCP server. A banner across the directory notes that new server submissions and listing changes are paused while the team reworks how it ingests and manages listings.
pricing See pricing
- + Live and browsable today, with a very large catalog (21,690+ servers claimed, updated daily)
- + Rich filtering and sorting — remote availability, classification, popularity, release date, and search
- + Standardized server.json files make installation and configuration details easy to compare
- + Popularity and visitor estimates help gauge which servers are actually used across the ecosystem
- + Free-to-browse directory pages and server profiles
- − New server submissions and listing changes are paused while the ingestion and management system is reworked
- − No published security scanning or sandbox benchmarking of listed servers — you get metadata, not measured performance
- − Server quality varies widely, from major platforms to niche single-purpose tools with minimal usage
- − Some listed servers carry their own per-call costs, which are surfaced on profiles but require you to evaluate each one yourself
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. It addresses a real gap: agent skills (SKILL.md instruction packages for Claude Code, Codex, Cursor, Gemini CLI and chat-app agents) are everywhere, but published research found roughly 26% of skills carry at least one vulnerability. SkillzBench's pillars are 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, then publishing the Lift in percentage points), Equip (one free MCP server plus per-agent loadouts where the agent only sees chosen skills and tools), and Stay current (tracking upstream changes, re-scanning, re-benchmarking, and notifying). A 0-100 trust score combines security, quality, license, maintenance and community signals, and money never touches scores. SkillzBench is pre-launch — the waitlist and newsletter are live, and the platform itself is being built.
pricing 0
- + Everything will be free — the directory, the MCP connector, benchmarks, and alerts
- + Designed around a documented problem: published research found ~26% of skills carry at least one vulnerability
- + Commit-SHA version pinning plus license checks and security scanning give a verifiable chain of trust
- + Sandbox benchmarks with and without each skill produce a concrete Lift figure, shown separately from ratings
- + Per-agent loadouts mean an agent only sees the skills and tools explicitly chosen for it
- + Advertising revenue never influences scores or listings
- − Pre-launch — the waitlist and newsletter are live, but the product itself is not yet public
- − No live user counts, skill counts, or published benchmark results exist yet
- − Focused on agent skills (SKILL.md packages) rather than the broader MCP server ecosystem PulseMCP catalogs
- − As a new platform, its trust scores and Lift figures will not have a track record until benchmarks are published
Verdict
Winner: none
These platforms solve different problems: PulseMCP is a live directory of MCP servers, while SkillzBench is a pre-launch directory and benchmarking service for AI agent skills. A small business can use PulseMCP today; SkillzBench's verification and benchmark value is still being built.
| Option | Overall | Breakdown |
|---|---|---|
| PulseMCP | 7 | : 9: 8: 7: 5: 5 |
| SkillzBench | 6 | : 3: 10: 8: 9: 7 |
What each platform actually catalogs
Trust and evidence compared
Practical fit for a small business
Which should you pick?
- :
- :
- :
- :
- :
- :
FAQ
Is PulseMCP free to use?
PulseMCP's server directory pages and individual server profiles are free to browse, and profiles label each server's cost — many are marked entirely free to use. Keep in mind that some listed MCP servers set their own per-call costs, which are shown on their profiles, so the free part is the directory itself rather than a guarantee about every listed server.
Is SkillzBench available yet?
SkillzBench is pre-launch and actively being built. The waitlist and newsletter are live, but the platform itself — the skills directory, the free MCP connector, the security scanning, and the sandbox benchmarks — is not yet public, and no live user counts, skill counts, or benchmark results have been published. Joining the waitlist is the way to get access as it rolls out.
What is the difference between an MCP server and an agent skill?
An MCP server is a connector that gives an AI agent access to external tools and data — for example GitHub, Slack, Notion, or a database. An agent skill is a SKILL.md instruction package that teaches an agent how to do a task well, used by tools like Claude Code, Codex, Cursor, and Gemini CLI. PulseMCP catalogs the former; SkillzBench catalogs the latter, so many teams will eventually use both.
Why does skill security matter for a small business?
Published research found that roughly 26% of agent skills carry at least one vulnerability. For a small business without a dedicated security team, that makes vetting before deployment essential. SkillzBench is being built to do that vetting automatically — pinning each version to a commit SHA, license-checking, security-scanning, and quality-judging every skill — so you do not have to audit SKILL.md packages by hand.
How does SkillzBench make money if everything is free?
SkillzBench earns revenue from advertising shown in and around its blog articles, and it never charges users for the directory, the MCP connector, benchmarks, or alerts. The platform states that ad money never influences scores or listings, and Lift results are always shown separately from trust ratings so measured performance stays distinct from opinion.
Can I still submit a new MCP server to PulseMCP?
Not currently. PulseMCP displays a notice across its directory that new server submissions and listing changes are paused while the team reworks how it ingests and manages listings, with a promise to reopen as soon as the rework is ready. Existing listings remain browsable and the directory continues to show daily updates for the servers already in it.