Skip to main content

A Python framework for building LLM-powered agents with modular state, tools, and MCP support.

Project description

Pygent

A Python framework for building LLM-powered agents with modular state management, composable operators, and LLM tool-calling integration (including MCP).

Features

  • Modular architecture — Agents, sessions, context, tools, and plans as composable modules
  • State management — Consistent save/load and serialization via PygentOperator
  • LLM-native — Messages and tools align with OpenAI-style APIs
  • MCP support — Use Model Context Protocol (SSE and stdio) tools via ToolManager
  • Toolkits — Built-in toolkits such as BashToolkits for bash shell access; TerminalToolkits and RestrictedTerminal remain as compatibility aliases.

Requirements

  • Python 3.11+

Installation

From the project root:

pip install -e .

Or with uv:

uv pip install -e .

Examples that read .env files use python-dotenv:

pip install -e ".[examples]"

Quick Start

from pygent.agent import BaseAgent
from pygent.context import BaseContext
from pygent.llm import AsyncRequestsClient
from pygent.message import UserMessage
from pygent.module.tool import ToolManager
from pygent.toolkits import BashToolkits

class MyAgent(BaseAgent):
    def __init__(self):
        super().__init__()
        self.llm = AsyncRequestsClient(
            base_url="https://api.openai.com/v1",
            api_key="YOUR_API_KEY",
            model_name="gpt-4",
        )
        self.tool_manager = ToolManager()
        bash = BashToolkits(session_id="quickstart", workspace_root=".")
        self.tool_manager.add_module("bash", bash)
        self.tool_manager.register_tools(bash.get_all_tools())

    async def run(self, user_input: str):
        context = BaseContext(system_prompt="You are a helpful assistant.")
        context.add_message(UserMessage(content=user_input))
        # ... tool loop and LLM calls (see examples/react_agent)
        return context

See examples/react_agent.py for a ReAct-style agent with tool calling, and examples/multi_session_agent.py for a session-aware variant.

Compatibility note: AsyncOpenAIClient is still exported as an alias of AsyncRequestsClient. TerminalToolkits and RestrictedTerminal(root_dir=...) still delegate to the bash toolkit for compatibility, while the registered tool name is bash.

Project Layout

  • pygent/ — Core package: agents, sessions, context, LLM clients, messages, modules, and toolkits
  • examples/ — Example agents (e.g. ReAct agent)
  • docs/Architecture and API documentation
  • scripts/ — Local smoke tests and utility scripts
  • tests/ — Pytest tests

Documentation

License

Apache-2.0. See LICENSE for details.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

pygent_ai-0.1.8.tar.gz (99.0 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

pygent_ai-0.1.8-py3-none-any.whl (80.0 kB view details)

Uploaded Python 3

File details

Details for the file pygent_ai-0.1.8.tar.gz.

File metadata

  • Download URL: pygent_ai-0.1.8.tar.gz
  • Upload date:
  • Size: 99.0 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for pygent_ai-0.1.8.tar.gz
Algorithm Hash digest
SHA256 b7adc093878d42eb92f8ba879f939ea4b3cccd6545255a843b8e3105112294a6
MD5 f70d2d1c252335e847c8410c50127784
BLAKE2b-256 d7996b0a5b1783dd2ec27a92e25afa221a30a31f746752ae06507145b6397cb9

See more details on using hashes here.

Provenance

The following attestation bundles were made for pygent_ai-0.1.8.tar.gz:

Publisher: python-publish.yml on pygent-ai/pygent

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file pygent_ai-0.1.8-py3-none-any.whl.

File metadata

  • Download URL: pygent_ai-0.1.8-py3-none-any.whl
  • Upload date:
  • Size: 80.0 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for pygent_ai-0.1.8-py3-none-any.whl
Algorithm Hash digest
SHA256 93f32c8aba29bdc3d13cf20384b99c03b021499cef7b9480273c6601e776dde7
MD5 65abf1762364ec6fa696a2a622863e5d
BLAKE2b-256 4e1c0720e32288c33e467bd7ddeac34369410c61be7660d332c0ad140eb70ebf

See more details on using hashes here.

Provenance

The following attestation bundles were made for pygent_ai-0.1.8-py3-none-any.whl:

Publisher: python-publish.yml on pygent-ai/pygent

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page