LLMContentBlock
API reference
A single piece of content within an LLM message.
Implements Node
arguments- The arguments passed to the tool, JSON-encoded (for TOOL_CALL kind).
callId- The unique ID of a tool call (for TOOL_CALL or TOOL_RESULT kinds).
data- Base64-encoded media bytes (for IMAGE, AUDIO, or DOCUMENT kinds).
errored- Whether the tool call resulted in an error (for TOOL_RESULT kind).
id- A unique identifier for this LLMContentBlock.
kind- The kind of content block, which determines the other populated fields.
mimeType- The media MIME type (for IMAGE, AUDIO, or DOCUMENT kinds).
signature- Provider-specific opaque data (e.g. Anthropic thinking signature). Preserve it when reconstructing a conversation.
text- Text content (for TEXT, THINKING, or TOOL_RESULT kinds).
toolName- The name of the tool called (for TOOL_CALL kind).
content- Ordered content returned by a tool, following any text (for TOOL_RESULT kind).
arguments: JSON!
The arguments passed to the tool, JSON-encoded (for TOOL_CALL kind).
callId: String!
The unique ID of a tool call (for TOOL_CALL or TOOL_RESULT kinds).
data: String!
Base64-encoded media bytes (for IMAGE, AUDIO, or DOCUMENT kinds).
errored: Boolean!
Whether the tool call resulted in an error (for TOOL_RESULT kind).
id: ID!
A unique identifier for this LLMContentBlock.
kind: LLMContentBlockKind!
The kind of content block, which determines the other populated fields.
mimeType: String!
The media MIME type (for IMAGE, AUDIO, or DOCUMENT kinds).
signature: String!
Provider-specific opaque data (e.g. Anthropic thinking signature). Preserve it when reconstructing a conversation.
text: String!
Text content (for TEXT, THINKING, or TOOL_RESULT kinds).
toolName: String!
The name of the tool called (for TOOL_CALL kind).
content: [LLMContentBlock!]!
Ordered content returned by a tool, following any text (for TOOL_RESULT kind).