Skip to main content
GET
/
v1
/
agent-configurations
/
{id}
/
evaluation-config
Get Evaluation Config
curl --request GET \
  --url https://api.example.com/v1/agent-configurations/{id}/evaluation-config
{
  "agent_api_key": "<string>",
  "agent_model_name": "<string>",
  "hub": "custom",
  "protocol": "chat_completions",
  "agent_url": "<string>",
  "agent_name": "<string>",
  "rate_limit_requests_per_minute": 100
}

Path Parameters

​
id
string<uuid>
required

Response

Successful Response

Configuration for Diamond evaluation (resolved from agent).

URL requirements depend on access level:

  • Black Box: agent_url is ALWAYS required (it's where the opaque agent lives)
  • Grey/White Box: agent_url is only required for non-standard hubs (custom LLM endpoints)

For standard hubs (OpenAI, Anthropic, etc.) in grey/white box mode, the SDK uses default endpoints so no URL is needed.

protocol is included in workflow params so the Diamond worker can normalize agent_url (e.g. .../v1/ for chat_completions) even when the stored workflow predates registry fixes.

​
agent_api_key
string
required
​
agent_model_name
string
required
​
hub
string
default:custom

Hub type for routing to correct SDK client

​
protocol
enum<string>
default:chat_completions

Agent protocol; used when normalizing agent_url for evaluation.

Available options:
chat_completions,
a2a
​
agent_url
string | null

Agent/LLM endpoint URL. Required for black box agents and custom hub grey/white box agents.

​
agent_name
string | null

Human-readable name of the agent (for reports and display)

​
rate_limit_requests_per_minute
integer
default:100

Rate limit for API requests per minute

Required range: x >= 1
Last modified on April 21, 2026