Actant
Actant is a durable Python agent runtime built on Temporal. Define agents and tools normally; Actant handles parallel tools, human approval, deferred work, subagents, suspension, and crash-safe continuation.
Actant is pre-1.0. Public APIs may change.
https://github.com/user-attachments/assets/021014df-4af6-47fe-8da9-cd6221008604
Why Actant
Agent tools become a distributed-systems problem when calls run in parallel, wait for people, or outlive a worker. Actant handles that orchestration:
- allowed tools execute concurrently;
- deferred tools pause without holding a worker;
- the next model turn waits for the complete tool group;
- approvals and nested-agent waits surface through the same API;
- Temporal recovers execution after process or worker failure;
- projection stores keep state easy for APIs and UIs to read.
flowchart TB
Turn["One agent turn emits A, B, and C"]
A["A: execute → completed"]
B["B: wait ··· human approves → completed"]
C["C: execute → completed"]
Barrier["Durable tool-group barrier"]
Next["Next agent turn"]
Turn --> A & B & C
A --> Barrier
B --> Barrier
C --> Barrier
Barrier --> Next
Read Why Actant? for the detailed guarantees and framework comparison.
Install
pip install actant
pip install "actant[openai]" # optional provider
pip install "actant[anthropic]" # optional provider
pip install "actant[gemini]" # optional provider
Start a local Temporal development server:
actant server start
The server stays attached so its logs and lifecycle remain visible. Pass
--detach only when you intentionally want it to run in the background.
Quickstart
This complete example hosts a worker, submits a message, and reads the persisted reply:
import asyncio
from contextlib import suppress
from uuid import uuid4
from temporalio.client import Client
from actant import AgentDefinition
from actant.llm.providers.fake import FakeLLM, FakeResponse
from actant.runtime import AgentRuntime
from actant.runtime.stores import InMemoryRuntimeStores
from actant.tools import ToolRegistry
agent = AgentDefinition(
id="assistant",
name="Assistant",
persona="You are useful.",
llm=FakeLLM([FakeResponse(text="Hello from Actant.", text_chunks=["Hello ", "from Actant."])]),
tools=ToolRegistry([]),
)
async def resolve_agent(agent_id: str, thread_id: str) -> AgentDefinition:
if agent_id != agent.id:
raise KeyError(agent_id)
return agent
async def main() -> None:
client = await Client.connect("localhost:7233")
runtime = AgentRuntime(
client=client, stores=InMemoryRuntimeStores(), resolve_agent=resolve_agent
)
polling = asyncio.create_task(runtime.run_worker())
try:
thread = runtime.thread(agent.id, uuid4())
workflow_id = await thread.send("hello")
await client.get_workflow_handle(workflow_id).result()
print((await thread.messages())[-1].content)
finally:
polling.cancel()
with suppress(asyncio.CancelledError):
await polling
asyncio.run(main())
# Hello from Actant.
thread.send() durably submits work and returns immediately. thread.events()
provides typed live deltas and lifecycle events; thread.messages() provides
the persisted reload path. Custom hooks and listeners remain available for
advanced worker-side callbacks.
The runtime has three write-side entry points:
thread = runtime.thread(agent.id, uuid4())
await thread.send("Start")
await thread.resolve(tool_call_id, approved=True)
await thread.cancel()
The equivalent runtime-level methods remain available when an application
already carries agent_id and thread_id separately. A thread handle also
exposes state(), messages(), waiting_tools(), and typed live events().
Use OpenAIProvider, AnthropicProvider, GeminiProvider, or QwenProvider
in place of FakeLLM. Actant never chooses a model ID for you.
Tools and approvals
Annotated functions become tools directly:
from actant import tool
@tool
async def weather(city: str) -> dict[str, str]:
"""Get the current weather for a city."""
return {"city": city, "forecast": "sunny"}
@tool(approval="Publish {title}?")
async def publish(title: str) -> dict[str, str]:
"""Publish an update."""
return {"published": title}
Register them with ToolRegistry([weather, publish]). Actant derives the JSON
schema from annotations. Approval tools enter the same durable AWAIT_HUMAN state as
advanced deferred tools and execute only after thread.resolve(..., approved=True).
Demo
The included FastAPI + React viewer demonstrates streaming, approvals, multiple-choice questions, mixed parallel tools, and nested subagents without an API key:
just demo-sync
just demo
Open http://localhost:5173.
Documentation
- Core concepts
- Runtime architecture
- Runtime and deployment
- Tools and approvals
- Pauses and deferred work
- Subagents
- Application coordinators
- Release process
Development
just sync
just test
just lint
just typecheck
just package
The justfile is repository-only. Installed users receive the actant CLI;
run actant server --help for local Temporal commands.
License
1 maintainer
Metadata
1 maintainer
Release files for actant 0.29.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| actant-0.29.0.tar.gz | 284.7 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| actant-0.29.0-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 462.1 kB
Release files / actant-0.29.0.tar.gz
| Download URL | actant-0.29.0.tar.gz |
|---|---|
| Size | 284.7 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
d5b6f3bf3cd4433ba3602d5545d4fda7b467fa1f90dbae811912f500381a7e03
|
|
BLAKE2b-256 checksum How to use checksums |
75dd6a4530523d896e84cfabad25b784b19a1c5e24812e851ff1cfef31544a91
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 1, 2026.
Transparency logRelease files / actant-0.29.0-py3-none-any.whl
| Download URL | actant-0.29.0-py3-none-any.whl |
|---|---|
| Size | 177.3 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
afbeb18e439eef99661e463290b6e143fca8662d5910b6e41216d606ab7314d4
|
|
BLAKE2b-256 checksum How to use checksums |
e8d903abb1acd8b6595c5bf72f2329e9ff505037c968186a3538e9e199c2d748
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 1, 2026.
Transparency log
