Agent Census

Snapshot of November 2, 2026. 61,058 listings. Free to download, no sign-up, no tracking.

An open dataset of publicly listed AI agents, bots, skills, and MCP servers, collected from the public indexes where these projects list themselves.

These are listings, not active agents. A row means an item was publicly listed at one of the sources below on the collection date. The census does not measure usage, quality, or whether a project is still running.

Composition

Kind	Listings	Share
MCP server	54,234	88.8%
Grok Bot template	5,264	8.6%
Agent skill (OpenClaw)	981	1.6%
Agent project	423	0.7%
Hugging Face Space	156	0.3%
Total	61,058

88.8 percent of the listings are MCP Registry entries, so the total mostly reflects that one catalog alongside the smaller sources below.

Sources

Source	What it lists	Listings
MCP Registry	MCP servers, official registry, active latest versions only	54,234
grokbots.best	Grok bot templates	5,230
ClawHub	OpenClaw skills from official publishers	981
e2b awesome-ai-agents (GitHub)	Agent projects	215
kyrolabs awesome-agents (GitHub)	Agent projects	208
Hugging Face	Spaces tagged as agents, with at least 2 likes	156
grokdex.net	Grok bot templates	34

Counts are after deduplication across sources. 113 cross-listings were merged, and the surviving row names the other index in its also_listed_at field.

Collection rules

Public sources only: public pages, feeds, and APIs.
robots.txt is checked per host before fetching and respected.
At most one request per second per host.
No personal data: author handles present in source payloads are not stored, and email addresses or phone-like numbers found in listing text are removed.
Refreshed by re-running the same collectors. New public listings at covered sources are included automatically.

Downloads

agents.csv: all 61,058 listings, 12 fields, 23.8 MB.
agents.json.gz: the same listings as JSON, gzip compressed to 5.1 MB (34.6 MB uncompressed). The JSON adds a meta block and a collector field per record.
Hugging Face dataset: snapshot CSVs with all 13 fields, including the prior snapshot of October 1, 2026.
Agent Directory API: search and per-listing lookup over the census. The API serves this same November 2, 2026 snapshot.

Fields

Field	Meaning
`id`	Census listing id
`name`	Project or server name as listed
`description`	Listing description as published by the source
`category`	Topic category assigned during the census
`kind`	One of the five kinds in the composition table
`platform`	Distribution form, for example "Open source / web"
`url`	The listing's own public URL
`source`	The index the listing was collected from
`source_url`	URL of that index
`also_listed_at`	Other indexes carrying the same listing, when known
`tags`	Topic tags, when the source provided them
`date_collected`	Collection date, 2026-11-02 for this snapshot

List-valued fields (also_listed_at, tags) are separated by semicolons in the CSV.

Corrections and removal

Listings are copied from public sources. If you own a listed item and want it corrected or removed, email abuse@tryoids.com with the subject "Agent Census removal" and include the listing name and source URL, or DM @oidsadmin on Oids. Removed rows go on a permanent do-not-list file so later refreshes do not re-add them.