A directory of the agentic AI stack: AI agents that do real work, MCP servers that connect them to your tools and data, and agent skills that extend what they can do. Every listing gets an indexable page, and pages link across the three layers, so an agent shows the MCP servers it works with and the skills that extend it. Search, categories, comparisons and alternatives run across all of it. Listing is free.
Every agent directory ignores MCP servers. Every MCP registry ignores agents. Skills have no directory at all beyond GitHub readmes. If you are building with agents today you are searching three places that do not know about each other.
AI Agents Listing indexes all three layers in one place and links them to each other. An agent's page shows the MCP servers it works with and the skills that extend it. An MCP server's page shows the tools it exposes and the skills that use it. Search and categories run across all three, and every listing has its own indexable page with a description, pricing, setup and links.
A few things I decided on purpose:
Rankings are engagement, not money. Paid tiers get placement and a label. The trending lists never sort by what somebody paid, because a list you can buy into is the list nobody uses.
Agents can read it too. There is an MCP endpoint that answers what is trending and what works with what, no login needed, and an llms.txt for the rest.
Solo project. It started from the Dirstarter directory boilerplate (Next.js and Postgres) and grew the three-layer model and the cross-links on top of it. Self-hosted. I read every submission myself.
I would rather have your criticism than your upvote. The thing I am least sure about is whether cross-linking three layers reads as useful or as clutter on a listing page, so tell me which it was for you.
AstraPixels
Hi Product Hunt 🔥.
Every agent directory ignores MCP servers. Every MCP registry ignores agents. Skills have no directory at all beyond GitHub readmes. If you are building with agents today you are searching three places that do not know about each other.
AI Agents Listing indexes all three layers in one place and links them to each other. An agent's page shows the MCP servers it works with and the skills that extend it. An MCP server's page shows the tools it exposes and the skills that use it. Search and categories run across all three, and every listing has its own indexable page with a description, pricing, setup and links.
A few things I decided on purpose:
Rankings are engagement, not money. Paid tiers get placement and a label. The trending lists never sort by what somebody paid, because a list you can buy into is the list nobody uses.
Agents can read it too. There is an MCP endpoint that answers what is trending and what works with what, no login needed, and an llms.txt for the rest.
Solo project. It started from the Dirstarter directory boilerplate (Next.js and Postgres) and grew the three-layer model and the cross-links on top of it. Self-hosted. I read every submission myself.
I would rather have your criticism than your upvote. The thing I am least sure about is whether cross-linking three layers reads as useful or as clutter on a listing page, so tell me which it was for you.
Thanks a lot
🔥❤️
vibecoder.date
This is what we are missing, a place that represents the whole ecosystem.
I'm keeping an eye on this. the landing page looks good, I see the vision here