curl https://api.60db.ai/agents/AGENT_ID/evals \
-H "Authorization: Bearer your-api-key"
Agent Reliability
Get Agent Evals
List the evals of an agent, and whether each one scores calls and has a golden check
GET
/
agents
/
{agentId}
/
evals
curl https://api.60db.ai/agents/AGENT_ID/evals \
-H "Authorization: Bearer your-api-key"
List the evals of an agent. There are six built-in evals, then the agent’s own custom rules. Use the names with Run golden eval.
The built-in evals are
Request
Headers
Path
Response
200 OK.{
"evals": [
{
"name": "hallucination", "kind": "pass_fail", "rule": "Did the agent avoid …", "source": "builtin", "enabled": true,
"golden": { "examples": 12, "agreed": 11, "false_passes": 0, "false_fails": 1, "unreliable": false, "judge_model": "judge-v1" }
},
{ "name": "out-of-scope", "kind": "pass_fail", "rule": "…", "source": "builtin", "enabled": false, "golden": null },
{ "name": "polite", "kind": "pass_fail", "rule": "Was the agent polite?", "source": "custom", "enabled": true, "golden": null }
]
}
| Field | Meaning |
|---|---|
source | builtin, or custom for the agent’s own rules |
enabled | Whether calls made as this agent are scored with this eval |
golden | The latest golden-set result on the Judge model in use now. null when it was never measured on that model. Custom rules have no golden set |
golden.unreliable | true when more than 10% of the passes were false. An eval like this cannot be newly enabled |
hallucination, out-of-scope, escalation-handled, required-disclosure, identity-before-data and no-pii-read-back.
Errors
| Status | Code | Meaning |
|---|---|---|
| 403 | POLICY_DENY | Your role cannot read agents, or the API key lacks the agents scope |
| 404 | AGENT_NOT_FOUND | No agent with this ID in your workspace |
| 409 | AGENT_NOT_SYNCED | Save the agent once, then try again |
Example
curl https://api.60db.ai/agents/AGENT_ID/evals \
-H "Authorization: Bearer your-api-key"