Method cheatsheet
The names you'll actually call. When unsure, list them:
[m for m in dir(client.atoms.agents) if not m.startswith("_")].
Agents — client.atoms.agents
| Do this |
Call |
| Create an agent |
create_agent(name=..., workflow_type="single_prompt", first_message=...) → .data is the id string |
| Get one |
get_agent(id=...) → .data |
| List |
list_agents() (wrap with helpers.as_page) |
| Archive |
archive_agent(id=...) |
| Duplicate |
duplicate_agent(id=...) |
| Create from a brief |
create_with_ai(description=...) |
| Call logs for an agent |
list_call_logs(id=...) |
| Do this |
Call |
| Add a transfer |
add_transfer_call(agent_id, number=..., transfer_type="cold_transfer"|"warm_transfer", on_hold_music=...) |
| Set/merge tools |
set_tools(agent_id, [...], replace=False, make_live=True) |
| Read live tools |
get_tools(agent_id) |
| Remove a tool |
remove_tool(agent_id, name) |
Do not set tools via create_agent/update_agent (no tools param) or the
legacy workflow doc (ignored on live calls).
Calls — client.atoms.calls
| Do this |
Call |
| Place outbound |
start_outbound_call(agent_id=..., phone_number=..., from_product_id=...) |
| Get one |
get(id="CALL-...") |
| List |
list(agent_ids=..., limit=..., call_types=...) → .data.logs |
| Search |
search(...) |
Versioning — client.atoms.agent_versioning_branches / ...revisions
Branch/revision model (v2). The v1 agent_versioning_drafts /
agent_versioning_versions are deprecated (return 409). For tools/prompt, prefer
AgentTools / the versioning helper.
Crew — smallestai.atoms.crew
| Thing |
Import |
| Node base |
from smallestai.atoms.crew.nodes import OutputCrewNode |
| Transfer event |
from smallestai.atoms.crew.events import SDKAgentTransferConversationEvent, TransferOption, TransferOptionType |
Waves (speech) — client.waves
| Do this |
Call |
| Text to speech (file) |
waves.synthesize_tts(...) / waves.synthesize_lightning(...) |
| Text to speech (SSE) |
waves.synthesize_sse_tts(...) |
| Text to speech (stream) |
waves.tts.connect(...) |
| Speech to text (file) |
waves.speech_to_text.transcribe(...) |
| Speech to text (stream) |
waves.speech_to_text.stream(...) |
| Voice cloning |
waves.create_voice_clone(...) / list_voice_clones() |