EvaluationSubject Fields

Field	Values	Description
`kind`	`custom_agent`	Always `custom_agent` for external agents
`displayName`	string	Human-readable name shown in the dashboard
`framework`	`raw_python`, `openai`, `anthropic`, `langchain`, `llamaindex`, `crewai`, `autogen`, `n8n`, `flowise`, other	Framework used
`runtime`	`local`, `ci`, `customer_hosted`	Where the agent runs
`version`	string	Optional version tag for the agent
`endpoint`	URL string	Optional, for HTTP-based agents

Return value from your agent function

Your callable can return any of:

Return type	Behavior
`str`	Used directly as the output text
`dict` with `"output"` key	Output text from `output`, rest stored as metadata
`EvaluationResult`	Full control - pass rating, justification, trace, timings

Security and privacy

The SDK automatically scrubs secrets from outputs and metadata before uploading:

sk-... API keys
Bearer tokens
Authorization headers
Password-like fields

Raw agent outputs, prompts, and CoT reasoning are never uploaded. Only the text response, metadata you explicitly include, and optional observable trace summaries.