All SDK methods return typed dataclasses with full IDE autocomplete.
Returned by agent.chat() and agent.achat().
| Field | Type | Description |
|---|
.content | str | LLM response text |
.model | str | Model used (e.g. gpt-4o-mini) |
.usage | Usage | Token usage |
.pipeline | PipelineResult | Pipeline execution details |
.token | CapabilityToken | Signed capability token |
.tool_calls | list[ToolCall] | Tool calls made by the LLM |
str(response) returns .content for convenience.
| Field | Type |
|---|
.prompt_tokens | int |
.completion_tokens | int |
.total_tokens | int |
| Field | Type | Description |
|---|
.blocked | bool | Whether the request was blocked |
.classification | str | Data classification (PUBLIC, INTERNAL, etc.) |
.steps | list | Steps that ran |
.flags | dict | Active governance flags |
| Field | Type | Description |
|---|
.token | str | Signed JWT |
.classification | str | Data classification |
.tools | list[str] | Allowed tools |
.ops | list[str] | Allowed operations |
.ttl | int | Time-to-live (seconds) |
| Field | Type |
|---|
.id | str |
.name | str |
.arguments | dict |
| Type | Returned by | Key fields |
|---|
HealthStatus | tp.health() | .status, .version, .storage |
AgentInfo | tp.agents() | .agent_id, .framework, .active |
PipelineInfo | tp.pipelines() | .pipeline_id, .steps, .agents |
AuditEntry | tp.audit() | .event_id, .event_type, .timestamp, .agent_id |
