Conversations and Transcripts

#Conversation IDs

Every run belongs to a conversation. Omit conversation_id to create one, then reuse the returned ID:

python

first = await agent.run("Show revenue by region")
 
second = await agent.run(
    "Now compare that with the prior quarter",
    conversation_id=first.conversation_id,
)

Conversation continuity is a bounded cold projection of completed runs, not a live session runtime. Daita projects at most:

8 prior runs;
40 messages; and
24,000 UTF-8 bytes.

Older context falls outside the projection rather than being summarized by another model call.

#Source Pinning

When a new conversation begins with an explicit source_id, Daita pins that source to the conversation:

python

first = await agent.run(
    "Summarize open orders",
    source_id=orders_source.id,
)
 
follow_up = await agent.run(
    "Group them by region",
    conversation_id=first.conversation_id,
)

An explicit conflicting source on a later turn is rejected. Use a new conversation to switch source scope.

#Inspect Conversation Runs

python

exists = await agent.conversation_exists(first.conversation_id)
runs = await agent.conversation_runs(first.conversation_id)
 
for item in runs:
    print(item.turn_index, item.transcript.run.message, item.result)

Each ConversationRun contains its zero-based turn index, exact Transcript, and terminal LoopExit when available.

#Per-Run Transcript

python

transcript = await agent.transcript(first.run_id)

The transcript contains only canonical messages from that run. Context projected from prior runs, memory, skills, and catalog facts is request context—not copied transcript history.

#Clear Conversations

python

deleted = await agent.clear_conversations()

Clearing conversations deletes transcripts, learning candidates, and review stamps. Separately approved memory, user profile, semantic annotations, and skills are preserved.

The bundled terminal requires confirmation:

bash

daita conversations clear atlas --yes