AgentCrush

Rankings/Service agents

Category · service agents

Service Agent Rankings

Service agents expose callable endpoints — they're functional, not economic, and not knowledge artefacts. They live on protocols (A2A, Agentverse, x402, ERC-8004, MCP) and earn ranking through actual adoption, source quality, ongoing activity, and how many service surfaces they're discoverable on.

v1.0 — LIVEmethodology: v1.1-service-forks

28 service agents tracked · 28 evidence-ranked. Currently sourced from A2A protocol agents (GitHub-discovered) and Agentverse (Fetch.ai). v1.1 will add ERC-8004 registry agents and Bazaar x402 endpoints as additional service surfaces.

Methodology

Composite is a weighted blend of six service-agent signals. Sub-scores are published; every weight is documented.

Adoption25%LIVEGitHub stars (A2A protocol agents) or interaction count (Agentverse). Log-scaled. Higher of the two wins.
Source Quality20%LIVEA2A signal_strength (0-100, based on stars + activity + topic match) OR Agentverse rating (0-5 → 0-100).
Activity Recency15%LIVETime since most recent GitHub push or Agentverse last-seen. Recent = high score, dormant = low.
Protocol Breadth15%LIVECount of declared protocols/topics (e.g. A2A, x402, MCP). Each protocol declared = +25 score.
Forks15%LIVEGitHub fork count, log-scaled. Forks measure active engagement (use/modify) vs passive starring. For service agents that expose code, this is a stronger adoption signal than stars.
Discourse / Social10%PLANNEDv1.1 will integrate X + Farcaster mention volume for service agents.

Evidence-ready rule

A service agent is evidence-ranked when at least 3 of 6 signals are present AND at least one is an adoption signal (GitHub stars > 0 OR Agentverse interactions > 0). Pure protocol-presence ≠ evidence-ranked — must show actual usage.

Tracked service agents (28)

Current coverage. Sub-scores visible per agent — methodology shows its work.

#1
A2Aevidence-ranked
GH: a2aproject/A2A · 23,775★ · 2,399 forks
798010010074⏳77
#2
a2a-pythonevidence-ranked
GH: a2aproject/a2a-python · 1,907★ · 436 forks
5910010010058⏳73
#3
evolverevidence-ranked
GH: EvoMap/evolver · 7,414★ · 756 forks
708010010063⏳73
#4
a2a-samplesevidence-ranked
GH: a2aproject/a2a-samples · 1,572★ · 670 forks
589010010062⏳72
#5
bitterbot-desktopevidence-ranked
GH: Bitterbot-AI/bitterbot-desktop · 1,604★ · 426 forks
588010010058⏳69
#6
yomoevidence-ranked
GH: yomorun/yomo · 1,903★ · 143 forks
598010010047⏳68
#7
awesome-a2aevidence-ranked
GH: ai-boost/awesome-a2a · 591★ · 113 forks
509010010045⏳67
#8
a2a-goevidence-ranked
GH: a2aproject/a2a-go · 373★ · 76 forks
469610010041⏳67
#9
a2a-javaevidence-ranked
GH: a2aproject/a2a-java · 415★ · 148 forks
477710010048⏳64
#10
commonlyevidence-ranked
GH: Team-Commonly/commonly · 509★ · 66 forks
497910010040⏳64
#11
EDDIevidence-ranked
GH: labsai/EDDI · 352★ · 113 forks
467610010045⏳63
#12
LLMTornadoevidence-ranked
GH: lofcz/LLMTornado · 605★ · 102 forks
50808010044⏳62
#13
sample-strands-agent-with-agentcoreevidence-ranked
GH: aws-samples/sample-strands-agent-with-agentcore · 157★ · 51 forks
407010010038⏳60
#14
awesome-a2aevidence-ranked
GH: pab1it0/awesome-a2a · 171★ · 38 forks
407010010035⏳59
#15
swivalevidence-ranked
GH: Swival/swival · 157★ · 15 forks
407010010026⏳58
#16
idun-agent-platformevidence-ranked
GH: Idun-Group/idun-agent-platform · 174★ · 9 forks
407010010021⏳57
#17
a2a-directoryevidence-ranked
GH: sing1ee/a2a-directory · 233★ · 27 forks
43738010031⏳57
#18
pilotprotocolevidence-ranked
GH: TeoSlayer/pilotprotocol · 104★ · 13 forks
366610010025⏳56
#19
story-shot-agentevidence-ranked
GH: neopen/story-shot-agent · 64★ · 11 forks
336310010023⏳54
#20
Handlerevidence-ranked
GH: alDuncanson/Handler · 40★ · 2 forks
29691001007⏳52
#21
awesome-agentsevidence-ranked
GH: Scottcjn/awesome-agents · 69★ · 16 forks
33638010026⏳52
#22
hyperterseevidence-ranked
GH: hyperterse/hyperterse · 81★ · 9 forks
34648010021⏳51
#23
A2AAppevidence-ranked
GH: tanaikech/A2AApp · 20★ · 4 forks
236410010013⏳51
#24
a2a-jsevidence-ranked
GH: a2aproject/a2a-js · 536★ · 138 forks
4979100⏳47⏳50
#25
a2a-adapterevidence-ranked
GH: hybroai/a2a-adapter · 52★ · 7 forks
31618010019⏳50
#26
markusevidence-ranked
GH: markus-global/markus · 44★ · 1 forks
30601001000⏳50
#27
dcp-aievidence-ranked
GH: dcp-ai-protocol/dcp-ai · 21★
2464100100⏳⏳49
#28
a2a-elixirevidence-ranked
GH: actioncard/a2a-elixir · 15★ · 3 forks
21628010010⏳46

v1.2 roadmap

v1.1 ships with concrete engagement signals (forks replaced the cross-protocol placeholder). v1.2 layers in cross-protocol presence + ecosystem reach.

  1. +1Cross-protocol presence (activate) — currently tracked in cross_protocol_presence table but unweighted in composite. Will activate as service agents start appearing on multiple surfaces beyond their source.
  2. +2ERC-8004 + Bazaar as service surfaces — 29K on-chain Base agents + 46K x402 endpoints. Treats paid endpoint as adoption proof.
  3. +3Contributor + commit-recency from GitHub — deeper code-health signals beyond raw forks.
All Rankings →Model Families →Tokenized →Labs →