II.
Benchmark JSON
Structured · livebenchmark:the-agent-company
TheAgentCompany json
Inspect the normalized record payload exactly as the atlas UI reads it.
{
"id": "benchmark:the-agent-company",
"_kind": "Benchmark",
"_file": "benchmarks/benchmarks/benchmarks-agentic.yaml",
"_cluster": "benchmarks",
"attributes": {
"displayName": "TheAgentCompany",
"homepageUrl": "https://the-agent-company.com/",
"kind": "full-stack",
"targetsKind": "AgentVersion",
"description": "CMU benchmark simulating a real software-company environment\n(Gitea, RocketChat, Plane, OwnCloud, etc.) where agents complete\nconsequential workplace tasks across tools.\n"
},
"outgoingEdges": [
{
"from": "benchmark:the-agent-company",
"to": "skill-area:multi-app-orchestration",
"kind": "covers",
"attributes": {
"attributes": {
"coverage": "full",
"weight": 0.6
}
}
},
{
"from": "benchmark:the-agent-company",
"to": "skill-area:bug-fixing-from-issues",
"kind": "covers",
"attributes": {
"attributes": {
"coverage": "partial",
"weight": 0.2
}
}
},
{
"from": "benchmark:the-agent-company",
"to": "domain:software-engineering",
"kind": "applies_to",
"attributes": {}
},
{
"from": "benchmark:the-agent-company",
"to": "domain:operations",
"kind": "applies_to",
"attributes": {}
}
],
"incomingEdges": []
}