Skip to main content

PraisonAI Plugins

PraisonAI Plugins is a foundational repository demonstrating how to extend the standard capabilities of the PraisonAI framework using the official Plugin Ecosystem.

Every design decision here revolves around our Agent-Centric Philosophy. You simply write standard python functions and register them. PraisonAI's architecture handles dynamically searching and integrating your plugins at startup without complex configuration files strings.

Features

This repository includes examples and templates covering all 6 Plugin Types categorized natively by the PraisonAI SDK:

  1. Hooks: Inject code into agent interactions like tracking API calls, displaying terminal logs, etc. (simple_logger, custom_tracer)
  2. Tools: Bundle functional python functions as tools that an agent can call automatically. (basic_tools)
  3. Guardrails: Validate outputs, ensure safety, intercept Prompt Injections, prevent PII escapes. (pii_guardrail)
  4. Policies: Set runtime execution contexts—for example blocking dangerous tool execution unconditionally. (strict_policy)
  5. Skills: Bundle higher-level instructions and techniques to transform how the agent "thinks". (researcher_skill)
  6. Integrations: Effortless connectors to 3rd party APIs, internal databases, or services like Slack, Jira. (slack_integration)

🚀 Quickstart for Non-Developers

Install

Getting started is designed to be frictionless for both beginners and experts:

# 1. Clone this package (the plugin repository)
git clone https://github.com/mervinpraison/praisonai-plugins.git
cd praisonai-plugins

# 2. Install the plugins to your global python environment
pip install -e .

That's it. PraisonAI automatically knows you installed these plugins. No messy config required.

Testing Agent Flow

By simply checking if the plugins are available, your standard PraisonAI Agent acts normally, but now runs with the enhanced logic in your Plugins. For example, your test_plugins_discovery.py handles:

from praisonaiagents import Agent

agent = Agent(name="Tester", instructions="Just repeat what I say.")
# This automatically executes all logic defined in plugins: Guardrails, Logging, etc
agent.start("Hello world!")

🛠️ Developer Guide: Building Custom Plugins

The PraisonAI SDK natively implements a powerful Protocol-Driven plugin system.

A Protocol refers to a strict interface specification defining WHAT functionality exists without touching HOW PraisonAI works internally. Your plugins implement hooks across an Agent's lifecycle without modifying core SDK files!

1. Write the Plugin

Inside the src/praisonai_plugins folder, you can create a standard python class mimicking this example Guardrail plugin (pii_guardrail.py):

from praisonaiagents.plugins.plugin import Plugin, PluginInfo, PluginHook

class PIIGuardrailPlugin(Plugin):
    """A protocol-driven plugin evaluating guardrails."""
    
    @property
    def info(self) -> PluginInfo:
        return PluginInfo(
            name="pii_guardrail",
            version="1.0.0",
            description="Blocks PII in responses.",
            author="PraisonAI",
            hooks=[PluginHook.AFTER_LLM] # Define when this plugin acts!
        )
        
    def after_llm(self, response: str, usage: dict) -> str:
        # Example Logic protecting SSN
        if "social security" in response.lower():
            return "[REDACTED BY GUARDRAIL]"
        return response

2. Export it via pyproject.toml

PraisonAI scans for [project.entry-points."praisonai.plugins"] to load classes automatically! In the pyproject.toml root file, export it:

[project.entry-points."praisonai.plugins"]
pii_guardrail = "praisonai_plugins.guardrails.pii_guardrail:PIIGuardrailPlugin"

3. Reinstall

pip install -e .

Now, every time any agent inside your Python environment receives a response containing "social security", this plugin intercepts it, evaluates it, and scrubs the PII.


Sandbox Backends (praisonai.sandbox)

Optional sandbox backends (e.g. Capsule for WebAssembly isolation) live in this repo and register via the praisonai.sandbox entry-point group — separate from lifecycle hooks (praisonai.plugins).

pip install praisonai-plugins[capsule]
from praisonaiagents.sandbox import SandboxManager, SandboxConfig

manager = SandboxManager(SandboxConfig.capsule())
result = manager.run_code("print('hello')", language="python")
Entry point group Purpose Example
praisonai.plugins Lifecycle hooks, guardrails, policies simple_logger, pii_guardrail
praisonai.sandbox Optional sandbox backends capsule

The PraisonAI Protocol Philosophy

Built with developers in mind, PraisonAI's architectural design features:

  • Zero Overhead: The underlying Core SDK limits itself to hooks and standard Dataclass instances. Lazy-imports mean plugin hooks effectively run with near-zero performance cost.
  • Safe Multi-Agent Design: Plugins don't hold shared state. If you orchestrate AgentFlow() loops, these plugins safely execute globally without thread-locking context bugs.
  • Protocol Extensibility: You're never boxed in. Whether it's the MemoryProtocol, ToolProtocol, or native events (before_agent, after_llm), the Plugin abstraction matches any scenario you face in building robust GenAI apps.

Metadata

Release files for praisonai-plugins 0.0.6

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for praisonai-plugins 0.0.6
File Size Uploaded Provenance Actions
praisonai_plugins-0.0.6.tar.gz 27.8 kB Without provenance Details for praisonai_plugins-0.0.6.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for praisonai-plugins 0.0.6
File Interpreter ABI Platform Actions
praisonai_plugins-0.0.6-py3-none-any.whl Python 3 none any Details for praisonai_plugins-0.0.6-py3-none-any.whl

Total release size: 55.0 kB

Release files / praisonai_plugins-0.0.6.tar.gz

Download URL praisonai_plugins-0.0.6.tar.gz
Size 27.8 kB
Tags Source
SHA-256 checksum
How to use checksums
a5c51a0ddd4ef0db398b31ecf5d9eb73c14c9ddf182748d0a493035164ee6ef9
BLAKE2b-256 checksum
How to use checksums
ab8806a92be0103ca09d11dffc1996ae356d0fa488469ff9109641a785c61a9a
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via uv/0.13.8 {"installer":{"name":"uv","version":"0.13.8","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Ubuntu","version":"26.04","id":"resolute","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":true}

Release files / praisonai_plugins-0.0.6-py3-none-any.whl

Download URL praisonai_plugins-0.0.6-py3-none-any.whl
Size 27.3 kB
Tags Python 3
SHA-256 checksum
How to use checksums
9d98ae55f3f6db3eba51089c274877b08bc838f6e4dbd3f4ca8c544e0aaa47a2
BLAKE2b-256 checksum
How to use checksums
732cb06b1302ed5909d8c15fd5d5447c280a7e5b31eb6ec4773e87fd8216427b
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via uv/0.13.8 {"installer":{"name":"uv","version":"0.13.8","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Ubuntu","version":"26.04","id":"resolute","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":true}

Security for praisonai-plugins 0.0.6

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.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page