Datasets:

AgentCrush
/

agents-index

license: cc-by-4.0
language:
  - en
tags:
  - ai-agents
  - agent-economy
  - rankings
  - mcp
  - erc-8004
  - x402
  - crewai
  - langchain
pretty_name: AgentCrush Agent Index
size_categories:
  - 1K<n<10K
task_categories:
  - text-classification
  - question-answering
configs:
  - config_name: agents
    data_files:
      - split: train
        path: data/agents.jsonl
  - config_name: evidence_ranked
    data_files:
      - split: train
        path: data/evidence-ranked.jsonl
  - config_name: snapshots_latest
    data_files:
      - split: train
        path: data/snapshots-latest.jsonl

AgentCrush Agent Index

Evidence-ranked index of the AI agent economy. Updated daily from agentcrush.xyz.

Overview

1,403 agents indexed across categories: developer tools, tokenized agents, service agents, model families
145 evidence-ranked with verified multi-signal scores
Updated: 2026-07-14

Configs

Config	Description	Rows
`agents`	All indexed agents with metadata	~1,403
`evidence_ranked`	Evidence-ranked tier only	~145
`snapshots_latest`	Most recent snapshot per agent	~1,403

Schema

Field	Type	Description
`handle`	string	Unique identifier (e.g. `crewai`, `aixbt_agent`)
`name`	string	Display name
`category`	string	`developer`
`tier`	string	`evidence_ranked`
`score`	float	0–100 composite score
`rank`	int	Rank within category
`weekly_delta`	int	Rank change vs previous week
`github_stars`	int	GitHub stars (if applicable)
`follower_count`	int	X/Farcaster follower count
`erc8004_verified`	bool	On-chain ERC-8004 identity verified
`x402_enabled`	bool	x402 payment endpoint active
`profile_url`	string	Full profile URL on agentcrush.xyz

Usage

from datasets import load_dataset

# All agents
ds = load_dataset("agentcrush/agents-index", "agents")

# Evidence-ranked only
top = load_dataset("agentcrush/agents-index", "evidence_ranked")

# Check a specific agent
df = ds["train"].to_pandas()
agent = df[df["handle"] == "crewai"].iloc[0]

# Filter by category
developer_agents = df[df["category"] == "developer"].sort_values("rank")

# Top movers this week
movers = df[df["weekly_delta"] > 5].sort_values("weekly_delta", ascending=False)

# ERC-8004 verified agents
verified = df[df["erc8004_verified"] == True]

Methodology

Rankings use per-category multi-signal scoring:

Developer: GitHub stars, forks, follower counts, activity signals
Tokenized: market cap, liquidity, holder count, momentum
Service: adoption, protocol presence, activity, forks
Model family: HF downloads, LMArena scores, derivatives, citations

Full methodology: agentcrush.xyz/methodology

License

CC-BY-4.0 Attribution: AgentCrush (agentcrush.xyz)

Citation

@misc{agentcrush2026,
  title={AgentCrush Agent Economy Index},
  author={AgentCrush},
  year={2026},
  url={https://agentcrush.xyz},
  note={Daily-updated dataset. agentcrush.xyz/methodology}
}