aicery-sdk 0.0.2

Python client for the Aicery AI agent runtime — runs, agents, and SSE streaming

aicery-sdk

Official Python client for the Aicery AI agent runtime.

About Aicery

Aicery is an open-source AI agent runtime — orchestration, tools, trace, and replay. It is the engine under your application, not a vertical SaaS product.

Agents are replaceable. Runtime is permanent.

The runtime runs LangGraph agents, executes tools safely, records step-by-step traces, and replays runs with frozen mocks. You bring UI, business rules, and data connectors; Aicery handles runs, providers, and observability.

• Repository: https://github.com/mrkkyatilla/aicery
• Documentation: https://aicery.dev/docs
• License: MIT

This package (aicery-sdk) is a thin HTTP client for the runtime API (/v1/runs, /v1/agents, SSE streaming). It does not include the runtime server — run the Aicery stack locally or point the client at your deployment.

Install

pip install aicery-sdk

Requires Python 3.11+.

Prerequisites

A running Aicery runtime (default: http://localhost:8000). From the main repo:

git clone https://github.com/mrkkyatilla/aicery.git
cd aicery
make install && make up && make migrate

Default API key for local dev: dev.

Quick start

from aicery_sdk import AiceryClient

client = AiceryClient(base_url="http://localhost:8000", api_key="dev")

run = client.create_run(agent_id="echo", input="hello", execute=True)
final = client.get_run(run.id)

print(final.status)       # completed
print(final.output_text)  # echo:hello

create_run may return status: pending immediately; call get_run for the final output.

Configuration

From aicery.yaml (recommended in projects):

client = AiceryClient.from_config("aicery.yaml")

From environment:

client = AiceryClient.from_env()
# API_KEY or AICERY_API_KEY (default: dev)
# AICERY_RUNTIME_URL (default: http://localhost:8000)

List agents

agents = client.list_agents()
for agent in agents:
    print(agent["id"], agent.get("name"))

Streaming

run = client.create_run(agent_id="research", input="Summarize README", execute=True)

for event in client.stream_run(run.id):
    if event.event == "token":
        print(event.data, end="", flush=True)

API surface (v0.0.x)

Method	Description
create_run(...)	Start or execute an agent run
get_run(run_id)	Fetch run status and output
list_agents()	List available agents
stream_run(run_id)	SSE token/step stream
resume_run(...)	HITL resume after approval
from_config(path)	Load client from aicery.yaml
from_env()	Load client from environment

Related

• Getting started guide
• Python SDK guide
• TypeScript SDK (npm)

Development

From the monorepo (editable install):

pip install -e ./sdk/python
pytest sdk/python/tests