Live data Live counts from public APIs — Hugging Face, GitHub, open directories.
Agent Scan · Updated weekly

Agents are being built now.
Agent Scan finds the best ones.

Agent Scan tracks verified agent counts and skills signals across business functions — weekly. Go-to-market is live now. Operations, Finance, Recruiting, and Software Development are coming next.

How the agent economy is structured

Three layers. One overwhelming market.

Understanding what exists — and why it's so hard to navigate — is the first step.

Layer 01
Skills

The smallest building block. A skill is a single reusable capability — write in a specific brand voice, format a newsletter, summarize a competitor's pricing page. Agents are assembled from skills the way a recipe is assembled from ingredients.

skills.sh90K+
SkillsMP96K+
LobeHub169K+
ClawHub / OpenClaw5,700+
ClaudeSkills.info658+
skills.shskillsmp.comlobehub.comclawhub
Layer 02
Where Agent Scan focuses
Agents

The workers. An agent takes a recurring job, runs it start to finish, and hands back a result — a newsletter draft, a competitive brief, five LinkedIn posts for your exec. Agent Scan tracks, vets, and surfaces these weekly across open marketplaces.

AI Agents Directory1,300+
AI Agents List750+
GitHub · new in 30 days6,000+
Hugging Face · agent-tagged2M+
New agents are published daily. Agent Scan tracks what's actually usable for GTM work — and publishes the findings weekly.
aiagentsdirectory.comaiagentslist.comgithub.comhuggingface.co
Layer 03
Applications

The finished products. Tools your team may already use — Monday.com, HubSpot, Zapier, Notion — are quietly becoming agent-powered. Powerful, but locked inside a subscription you already own.

Monday.comHubSpotZapier AgentsNotion AISalesforceMicrosoft 365+ hundreds more
These are walled gardens. Agent Scan works with open marketplaces only — so you can hire the best agent, not the one that happens to be included in your subscription.
context onlyopen markets only
Scale of the ecosystem

Why navigation requires a machine.

The full open AI ecosystem across public registries. Agent Scan's 1,466 verified GTM agents are a curated, filtered slice. That filter is the work.

2,000,000+
AI models & agents — open registries
huggingface.co · live API · ecosystem scale indicator
360K+
Total skills — open registries
skills.sh · skillsmp · lobehub · clawhub
6,000+
New AI agent repos — last 30 days
github.com · live API · topic:ai-agent
What Agent Scan does with this
Agent Scan filters this ecosystem down to what is actually callable, actually tested, and actually built for GTM workflows. The Hugging Face counter includes everything in the open registry — it is an ecosystem scale indicator. Agent Scan's Layer 1 count is a small, curated slice — filtered for open marketplaces only. The gap between 2M+ and 1,466 is the work of the scan.
Agent Scan · Updated weekly · June 2026

Two layers of intelligence.
One picture of the market.

Agent Count is what's callable today. Skills Signal is what's being built for tomorrow. Read them together and you can see where the market is heading before it gets there.

Layer 1 · Agent Count
Verified GTM agents —
callable today
Discrete agents across open marketplaces — agent.ai, Relevance AI, agentalent.ai, agentplace.ai — filtered for GTM use cases. Skills, MCP connectors, and embedded SaaS AI are not counted.
Updates weekly · ▲ = change this week   mo = change this month
✍️Content
312▲+4 wk ▲+32 mo
🎯Prospecting
287▲+6 wk ▲+27 mo
📣Social & LinkedIn
241▲+5 wk ▲+21 mo
🕵️Comp. Intel
198▲+5 wk ▲+18 mo
🔍SEO
178▲+4 wk ▲+16 mo
📧Email
156▲+4 wk ▲+16 mo
💰Revenue Ops
94▲+3 wk ▲+10 mo
agent.airelevanceai.comagentalent.aiagentplace.ai
Layer 2 · Skills Signal
Where builders are
experimenting today
Skills volume from skills.sh by GTM category — a leading indicator. Skills are recipes. Agents are cooks. skills.sh is the recipe library. Agent Scan tracks which recipes are being copied most — agents follow in 60 to 180 days.
Updates weekly · ▲ = change this week   mo = change this month
✍️Content
8,240▲+100 wk ▲+440 mo
🎯Prospecting
6,180▲+80 wk ▲+380 mo
📣Social & LinkedIn
5,340▲+80 wk ▲+340 mo
🕵️Comp. Intel
4,920▲+60 wk ▲+320 mo
🔍SEO
4,130▲+50 wk ▲+250 mo
📧Email
3,860▲+40 wk ▲+240 mo
💰Revenue Ops
2,180▲+50 wk ▲+160 mo
skills.shfiltered for gtm use cases

Where the next wave is forming

High skills volume + low agent count = the gap. These three categories show the strongest signal that new agents are incoming in the next 60–180 days.

🔴 Signal #1 · Highest gap ratio
Competitive Intel
198 agents · 4,920 skills · ratio 1:24.8
Skills volume is 25× the agent count. Builders are writing the playbook — agents are 60–90 days behind. Watch this category closely.
🟡 Signal #2 · Fastest-moving
Email
156 agents · 3,860 skills · ratio 1:24.7
Underserved relative to skills volume. Personalization is driving builder activity faster than agent supply is responding.
🟢 Signal #3 · Emerging early
Revenue Ops
94 agents · 2,180 skills · ratio 1:23.2
Smallest category — but identical gap ratio to the others. Whoever builds the right GTM ops agent here is early.
The opportunity map · Agent Scan

Where the next wave of agents
is coming — and how fast.

Each bubble is a GTM category. Position shows skills vs. agents. Bubble size = gap ratio — bigger means more builder activity relative to available agents. Below the line = behind on agents — those categories are where the next wave is forming.

Illustrative · updates weekly
as Agent Scan data is collected
Content
Prospecting
Social & LinkedIn
Competitive Intel
SEO
Email
Revenue Ops
Supply/demand parity line
Growing in your market right now

Models, tools & agents built for your GTM work.

Counts across open registries — not all production-ready. Agent Scan tracks and vets what's actually usable for specific GTM workflows.

✍️
Content
47,200
models, tools & agents
🎯
Prospecting
38,600
models, tools & agents
🕵️
Competitive Intel
31,200
models, tools & agents
📣
Social & LinkedIn
29,100
models, tools & agents
📧
Email
24,800
models, tools & agents
🔍
SEO
31,800
models, tools & agents
💰
Revenue Ops
18,400
models, tools & agents
New activity · open registries 0 added this session
Watching registries for new activity…
🔒 Coming soon

Agent Scan is expanding beyond go-to-market.

Go-to-market is live. These business functions are next — vote for the one you want to see first and we'll reach out when it launches.

Coming soon
⚙️
Operations
Process automation, workflow agents, ops intelligence — the function where agent ROI shows up fastest.
Coming soon
🎯
Recruiting
Sourcing, screening, outreach, and pipeline agents — the hiring stack is getting rebuilt with AI.
Coming soon
💻
Software Dev
Code generation, review, testing, and deployment agents — the category with the most active builder community by far.
Coming soon
💼
Finance
Reporting, forecasting, reconciliation, and FP&A agents — finance teams are early adopters of structured-output agents.
Agent Watch Review · Updated weekly

This week's featured agent.

One agent. One honest assessment. Updated each week with a new pick from the Agent Scan index.

View all reviews →
✍️ Content · Newsletter
PostCraft AI
postcraft.ai · agent.ai registry
Cost per run $4.00
Suggested cadence Weekly
Registry rating 4.7 ★ · 1.2k runs
Week of June 2, 2026
This week's assessment

Strong fit for teams running a
biweekly B2B newsletter.

PostCraft AI takes a Slack-channel brief and returns a full newsletter draft — intro, three sections, CTA, and a subject line. The output format is clean enough to paste directly into Beehiiv or Substack without heavy editing. The agent's handling of brand voice is the strongest we've tested in this category: provide a two-paragraph voice guide and it holds it consistently across runs.

The weakness is research depth. It pulls from provided URLs but won't proactively surface new sources. For newsletters that require original research or a strong point of view, treat it as a first-draft engine, not a replacement for editorial judgment. For curation-heavy or roundup formats, it's close to production-ready.

Strong fit
Biweekly or weekly newsletter · B2B SaaS · Curation or commentary format
Read full review → Next review drops Monday

Stay ahead of
a market that moves weekly.

Agent Watch is the weekly publication that turns Agent Scan data into readable intelligence for GTM leaders — what's new, what's moving, and what to act on.

Read Agent Watch on Substack → Get on the list
On these numbers: Agent Scan tracks two layers of the open AI agent ecosystem weekly. Layer 1 (Agent Count) counts discrete, callable agents across open marketplaces — agent.ai, Relevance AI, agentalent.ai, agentplace.ai — filtered for GTM use cases. Skills, MCP connectors, and SaaS features with embedded AI are not counted as agents. Layer 2 (Skills Signal) tracks task-relevant skills indexed on skills.sh by GTM category, used as a leading indicator of agent development. Walled gardens (Microsoft Copilot Store, Salesforce AgentExchange, Google Agentspace, AWS) are permanently excluded — they are not open marketplaces. The Hugging Face counter reflects live API data from HF's public endpoint and includes all models, tools, and agent-tagged items across the registry — it is an ecosystem scale indicator, not an Agent Scan count. Baseline numbers are set at Week 0 and updated weekly. Nothing on this page is fabricated or extrapolated beyond what the sources support.