LobeHub vs SkillzBench: Which Agent Skills Platform Should You Trust?
LobeHub is an agent platform: it runs AI agents 24/7 in the cloud, with a Skills Marketplace of 334,199+ skills and an MCP marketplace of 111,995+ servers, on a free tier plus paid plans from $12.90/month. SkillzBench is a free, independent directory that verifies, security-scans, and benchmarks agent skills, then connects them to your agents through a free MCP server. If you want a place to run agents, LobeHub fits; if you want to know which skills are safe and actually help before you install them, SkillzBench is the trust layer, and the two can work together.
TL;DR
- LobeHub is an agent operations platform: its Chief Agent Operator hires, schedules, and reports on AI agents running 24/7 in the cloud.
- LobeHub's Skills Marketplace lists 334,199+ skills and its MCP marketplace lists 111,995+ servers, but the platform does not publish per-skill security scans or benchmark results.
- SkillzBench is a free directory that verifies, security-scans, license-checks, and benchmarks agent skills, publishing a Lift figure for each skill tested with and without it in a sandbox.
- SkillzBench pins every skill version to a commit SHA and scores trust 0-100 from security, quality, license, maintenance, and community signals; money never touches scores.
- LobeHub's free plan includes 500,000 credits per month; paid plans run $12.90, $24.90, and $49.90 per month in the site's plan comparison table.
- SkillzBench charges nothing for anything: the directory, the MCP connector, benchmarks, and alerts are all free, funded by blog advertising.
- The two are complementary: run agents on LobeHub, and use SkillzBench's free MCP and loadouts to decide which skills those agents should get.
Head to head
| Dimension | LobeHub | SkillzBench |
|---|---|---|
| Pricing | Free tier (500,000 credits/mo); paid from $12.90/mo up to $49.90/mo | Free for everything; funded by blog advertising 🏆 |
| Core purpose | Run and orchestrate AI agents 24/7 in the cloud | Verify, benchmark, and connect agent skills |
| Skill safety | Marketplace listings without published per-skill security scans or benchmarks | Every skill pinned to a commit SHA, security-scanned, license-checked, and sandbox-benchmarked 🏆 |
| Skill breadth | 334,199+ skills in its Skills Marketplace 🏆 | Hundreds of skills crawled and deduped from GitHub; see skillzbench.com for current numbers |
| Ease of use | Browser-based cloud with guided plan and provider setup; usage dashboards per agent | Toggle skills per agent in loadouts; connect via one free MCP server |
| Integrations | Dozens of model providers (OpenAI, Anthropic, Google, Ollama, LM Studio, and more); Slack and Discord reporting 🏆 | One free MCP server connects any agent; per-agent loadouts |
| Support | Email and community forum on Free/Starter; priority email on Premium; priority chat and email on Ultimate 🏆 | Self-serve directory, benchmarks, and documentation |
| Setup | Sign up, pick a plan, configure model providers per agent | Connect the free MCP server and toggle skills per agent 🏆 |
The contenders
LobeHub positions itself as a 'Chief Agent Operator': a platform that organizes your AI agents into 7x24 operation, hiring, scheduling, and reporting on an entire agent team so you don't have to stay online. It combines an Agent Marketplace for building agents with full control over their skills, a Skills Marketplace listing 334,199+ skills, and an MCP marketplace with 111,995+ servers. Agents can run long-horizon tasks in parallel in the cloud and report through channels like Slack and Discord. On the model side, LobeHub supports dozens of providers, including OpenAI, Anthropic, Google, DeepSeek, Qwen, OpenRouter, and local runtimes like Ollama and LM Studio, with per-agent model choice. It also offers Pages for multi-agent writing, scheduling, projects, shared team workspaces, and personal memory that learns how you work. LobeHub began as an open-source LLM web UI (the GitHub project has around 59K stars) and remains free to self-host, while LobeHub Cloud adds hosted convenience, credits, storage, and an enterprise edition for private deployment. For a small business, it's a serious option if you want agents doing ongoing work, not just chatting.
pricing 12.9/month · free trial
- + Free tier with 500,000 credits per month lets you test the platform without paying
- + Enormous ecosystem: 334,199+ skills and 111,995+ MCP servers listed in its marketplaces
- + Runs agents autonomously 24/7 and reports through Slack and Discord, so work continues while you're offline
- + Dozens of model providers supported, including local inference via Ollama and LM Studio to avoid cloud API charges
- + Transparent usage tracking down to per-agent cost and token breakdowns
- + Open-source roots with a self-hosted option and an enterprise edition for private deployment
- − The pricing page shows conflicting figures for the same plans ($9.90/$19.90/$39.90 in one section, $12.90/$24.90/$49.90 in the plan comparison), which makes budgeting confusing
- − No published security scanning or benchmark results for the skills in its marketplace, and published research has found roughly 26% of agent skills carry at least one vulnerability
- − You can't downgrade directly; you must cancel, wait for the plan to expire, then re-subscribe at the lower tier
- − Premium features like Agent Memory and early access to top models are gated behind paid tiers
- − Heavy credit-based pricing means estimating message costs per model takes some work
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 are SKILL.md instruction packages for Claude Code, Codex, Cursor, Gemini CLI, and chat-app agents, and they've spread fast without much quality control: published research found roughly 26% of skills carry at least one vulnerability. SkillzBench tackles that with five pillars. Discover continuously crawls and dedupes skills from GitHub. Verify pins every version to a commit SHA, then validates, license-checks, security-scans, and quality-judges it. Benchmark tests every skill in a sandbox with the same model, with and without the skill, and publishes the Lift in percentage points. Equip gives you one free MCP server plus per-agent loadouts, so you toggle skills and tools per agent and the agent only sees what you chose. And Stay current tracks upstream changes, re-scans, re-benchmarks, and notifies you. A 0-100 trust score combines security, quality, license, maintenance, and community signals, Lift is always shown separately, and money never touches scores. Everything on the platform is free; revenue comes from advertising around blog articles.
pricing 0
- + Everything is free: the directory, MCP connector, benchmarks, and alerts
- + Every skill is pinned to a commit SHA and security-scanned, addressing the ~26% vulnerability rate found in published research
- + Lift benchmarks show the measurable effect of a skill, tested with the same model with and without it
- + Per-agent loadouts give precise control over what each agent can access
- + Ongoing monitoring means a skill that turns risky upstream gets re-scanned and you get notified
- + Trust scores are independent; advertising revenue never influences scores or listings
- − It doesn't run your agents; you need an agent platform or coding agent (Claude Code, Codex, Cursor, Gemini CLI) to use the skills it verifies
- − No dedicated support tiers; it's a self-serve directory and tool
- − Coverage is limited to skills it has crawled and verified from GitHub, so niche skills may not be listed yet
- − It won't host conversations, memory, or team workspaces the way a full agent platform does
Verdict
Winner: none
These two solve different problems. LobeHub is a paid agent platform for running and orchestrating AI agents; SkillzBench is a free verification and benchmarking layer for the agent skills those agents use, so most small businesses will want LobeHub for execution and SkillzBench for trust.
| Option | Overall | Breakdown |
|---|---|---|
| LobeHub | 8.2 | : 7: 8: 9: 6: 9: 4 |
| SkillzBench | 8.5 | : 10: 8: 8: 6: 8: 10 |
Pricing compared
Skill trust and safety
Where each one fits in your workflow
Which should you pick?
- :
- :
- :
- :
- :
FAQ
Can I use SkillzBench with LobeHub?
Yes. SkillzBench provides one free MCP server that connects any agent, and per-agent loadouts let you toggle exactly which skills and tools each agent sees. Since LobeHub supports MCP integrations (its own marketplace lists 111,995+ MCP servers), you can use SkillzBench's verified skills and trust scores to decide what to equip your LobeHub agents with, then connect them through the free MCP server.
How much does LobeHub cost for a small team?
LobeHub Cloud has a free plan with 500,000 credits per month, then paid plans listed at $12.90/month (Starter), $24.90/month (Premium), and $49.90/month (Ultimate) in the plan comparison on lobehub.com/pricing, with annual billing at $118.80, $238.80, and $478.80 respectively. Note the same page shows $9.90/$19.90/$39.90 in another section, so verify the price at checkout. Credits map to tokens, and each model consumes credits at a different rate.
Why should I worry about agent skill safety?
Agent skills are SKILL.md instruction packages that tell agents like Claude Code, Codex, Cursor, and Gemini CLI how to behave, and they're often installed without scrutiny. Published research found roughly 26% of skills carry at least one vulnerability. SkillzBench addresses this by pinning every skill version to a commit SHA, running security scans and license checks, and publishing a 0-100 trust score plus sandbox benchmark Lift results, all free.
Is SkillzBench really free, and how does it make money?
Yes, everything on SkillzBench is free: the skills directory, the MCP connector, benchmark results, and change alerts. The platform earns revenue from advertising shown in and around its blog articles, and that ad money never influences trust scores or listings. Lift results are always shown separately from ratings, so commercial incentives don't touch the evidence you see.
What is Lift in SkillzBench's benchmarks?
Lift is the measured performance change from adding a skill. SkillzBench tests every skill in a sandbox with the same model, once with the skill and once without, and publishes the difference in percentage points. It's always displayed separately from trust ratings, so you can see both how safe a skill is and how much it actually helps before you install it.
Do I need both platforms, or just one?
It depends on what you're doing. If you only need to run agents on ongoing tasks, LobeHub does that and starts free with 500,000 credits per month. If you only need to vet and connect skills to agents you already run (Claude Code, Codex, Cursor, Gemini CLI, or chat-app agents), SkillzBench covers that for free. Most small businesses adopting agents will benefit from both: a platform to execute and SkillzBench to decide which skills those agents can use.