rw_project_ids — this is the project where extracted facts and memory are stored.
Permission required:
workspace:conv_create on the workspace, plus one set per participant you name — on each project in rw_project_ids: project:mem_add, project:doc_add, project:mem_search and project:doc_search; on each actor in actor_ids: workspace_actor:fact_add and workspace_actor:fact_search; on each agent in agent_ids: workspace_agent:fact_add and workspace_agent:fact_search. A conversation both writes memories and reads them back, which is why each participant needs the add and the search half.Path Parameters
string
required
Workspace identifier
Request Body
string
required
Your own identifier for this conversation — the one it already has in your system. Must be unique within your tenant. Pass
by_custom_id=true on later requests to address the conversation by it instead of by id. See Your own identifiers.string
required
Conversation kind:
DIRECT (one-on-one) or GROUP (multi-participant).array
required
Read-write project(s) the conversation reads context from and writes facts to. Exactly one project ID is required.
array
required
Actors participating in this conversation. At least one participant is required across
actor_ids and agent_ids: a participant is what a message is attributed to and whose memory the conversation writes to, so a conversation without one can neither tell its participants apart nor build anyone’s memory.array of strings
Agents taking part in the conversation.
string
Display name for the conversation
object
Arbitrary key-value metadata to attach to the conversation
Response
object