AgentKit
A lightweight Python framework for building multi-agent systems with session management, message routing, state persistence, and lock management.
Installation
pip install agentkit-py
Features
- Session Management - Track agent sessions with ownership, channels, and metadata
- Message Routing - Send messages between agents with pub/sub support
- State Persistence - Key-value store with namespaces and atomic operations
- Lock Management - Distributed locking with TTL and cross-process support
Quick Start
from agentkit import SessionManager, MessageBus, StateStore, LockManager
from pathlib import Path
# Create managers with persistence
data_dir = Path("./agent_data")
sessions = SessionManager(data_dir / "sessions")
messages = MessageBus(data_dir / "messages")
state = StateStore(data_dir / "state.json")
locks = LockManager(data_dir / "locks")
# Create a session
session = sessions.create(
agent_id="agent-1",
channel="telegram",
owner="user-123",
)
# Send messages between agents
messages.send(
from_agent="agent-1",
to_agent="agent-2",
content="Hello from agent 1!",
)
# Receive messages
msg = messages.receive("agent-2")
if msg:
print(f"Got: {msg.content}")
# Persist state
state.set("last_run", "2024-01-15", namespace="agent-1")
state.increment("run_count", namespace="agent-1")
# Lock shared resources
with locks.hold("shared-resource", owner="agent-1", ttl_seconds=30) as lock:
if lock:
# Do exclusive work
pass
Session Management
from agentkit import SessionManager
sessions = SessionManager("./sessions")
# Create session
session = sessions.create(
agent_id="my-agent",
channel="slack",
owner="user-456",
metadata={"workspace": "acme"}
)
# Find sessions
active = sessions.find(agent_id="my-agent", status="active")
# Update session
session.metadata["last_action"] = "processed_message"
sessions.update(session)
# Cleanup stale sessions (inactive > 1 hour)
cleaned = sessions.cleanup_stale(max_age_seconds=3600)
Message Routing
from agentkit import MessageBus
bus = MessageBus("./messages")
# Send a message
msg = bus.send(
from_agent="coordinator",
to_agent="worker-1",
content="Process this task",
metadata={"task_id": "abc123"}
)
# Subscribe to messages
def handle_message(msg):
print(f"Received: {msg.content}")
bus.subscribe("worker-1", handle_message)
# Broadcast to all agents
bus.broadcast(
from_agent="coordinator",
content="System maintenance in 5 minutes"
)
# Check pending messages
count = bus.pending_count("worker-1")
State Persistence
from agentkit import StateStore
store = StateStore("./state.json")
# Basic operations
store.set("config", {"timeout": 30}, namespace="agent-1")
config = store.get("config", namespace="agent-1")
# Atomic increment
count = store.increment("requests", namespace="agent-1")
# Append to list with max length
store.append("recent_errors", "timeout error",
namespace="agent-1", max_length=100)
# Get with metadata
info = store.get_with_metadata("config", namespace="agent-1")
# {'value': {...}, 'updated_at': 1705334400.0}
Lock Management
from agentkit import LockManager
locks = LockManager("./locks")
# Acquire lock with TTL
lock = locks.acquire(
resource="database-connection",
owner="agent-1",
ttl_seconds=60
)
if lock:
try:
# Do exclusive work
pass
finally:
locks.release("database-connection", "agent-1")
# Context manager (auto-release)
with locks.hold("resource", "agent-1", ttl_seconds=30) as lock:
if lock:
# Work with lock held
pass
# Wait for lock
lock = locks.acquire(
resource="busy-resource",
owner="agent-1",
wait=True,
wait_timeout=10.0
)
# Force release stuck lock
locks.force_release("stuck-resource")
In-Memory Mode
All components work without persistence (in-memory only):
# No storage_dir = in-memory only
sessions = SessionManager()
messages = MessageBus()
state = StateStore()
locks = LockManager()
For AI Agents
See SKILL.md for agent-optimized documentation with command patterns and examples.
License
MIT
Metadata
Release files for agentkit-py 0.1.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| agentkit_py-0.1.0.tar.gz | 10.5 kB | Without provenance | Details for agentkit_py-0.1.0.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| agentkit_py-0.1.0-py3-none-any.whl | Python 3 | none | any | Details for agentkit_py-0.1.0-py3-none-any.whl |
Total release size: 23.2 kB
Release files / agentkit_py-0.1.0.tar.gz
| Download URL | agentkit_py-0.1.0.tar.gz |
|---|---|
| Size | 10.5 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
11e30e6ae342d91929e739d0e8dd3f76283808d92f6a99fe6f1474e6d8a93d98
|
|
BLAKE2b-256 checksum How to use checksums |
0971beba8283e409ab112c053fff48538f11ac7eb7ff237e3437028bbec427ca
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.14.2
|
Release files / agentkit_py-0.1.0-py3-none-any.whl
| Download URL | agentkit_py-0.1.0-py3-none-any.whl |
|---|---|
| Size | 12.7 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
d12ff811f2b42cf514c293cdae9212e03ac02e54099ead792aa67940312c0de9
|
|
BLAKE2b-256 checksum How to use checksums |
b742ae78d3978522e1fdef617dd97e1aea54c2ec2afdff7876d75103510df953
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.14.2
|
Security for agentkit-py 0.1.0
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.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.