Why we publish the funnel
The first post on the Agenstry blog — why a registry that hides its discovery funnel can't be trusted, and what we publish instead.
Most agent registries publish a count. 11,738 agents indexed. It looks like abundance, but it's actually opacity: you have no way to tell how many of those agents respond, how many serve a valid card, how many are dead.
We publish the funnel instead:
- {discovered} discovered across federated sources
- {responded} responded to a probe in the last 24 hours
- {actually work} returned a valid, schema-conforming card
Every step is reproducible from public data. If our number disagrees with yours, one of us has a bug — and we'd like to know.
What you'll find on this blog
Three kinds of posts, no announcements:
- Methodology notes. When we change how a criterion is scored, we'll write up why, what the failure mode was, and what we expect to see in the next month of data.
- Field reports. Things we learn from running a probe against ~12k agents that the spec authors couldn't have predicted. Most of them are the not-fun kind.
- Interop receipts. Side-by-side card diffs when A2A or MCP servers move between versions, with a fixed-width breakdown of what changed and what broke.
If you came here for a "we raised a Series A" post, you're in the wrong place. We have a 9-criterion methodology to maintain.
← Back to blog
Agenstry