Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.perplexity.ai/llms.txt

Use this file to discover all available pages before exploring further.

​
Overview

AG2 ships a PerplexitySearchToolkit that wraps the Perplexity Search API as a native AG2 tool. The tool returns ranked web results — title, URL, snippet, and date — without generating an LLM answer, which keeps token usage minimal and lets the calling agent decide how to consume them.
AG2 (formerly AutoGen) is an open-source framework for building multi-agent conversational AI systems. Learn more at ag2.ai.

​
Installation

pip install "ag2[perplexity]"

​
API Key Setup

Set your Perplexity API key as an environment variable, or pass it to the toolkit constructor: 
export PERPLEXITY_API_KEY="your_api_key_here"

Get API Key

Generate your API key from the Perplexity dashboard.

​
Quick Start

Register PerplexitySearchToolkit with an AG2 Agent: 
from autogen.beta import Agent
from autogen.beta.tools.search import PerplexitySearchToolkit

toolkit = PerplexitySearchToolkit()
search = toolkit.search()

agent = Agent(
    "researcher",
    prompt="Answer using fresh information from the web.",
    tools=[search],
)

await agent.ask("What were the top tech announcements this week?")
The tool calls the Perplexity Search API under the hood and returns structured PerplexitySearchResult objects.

​
Configuration

All search parameters are passed to toolkit.search() and forwarded to the Search API: 
from autogen.beta.tools.search import PerplexitySearchToolkit

toolkit = PerplexitySearchToolkit()
search = toolkit.search(
    max_results=5,
    max_tokens_per_page=512,
    search_domain_filter=["arxiv.org", "-medium.com"],  # prefix with '-' to exclude
    search_recency_filter="week",                         # hour | day | week | month | year
    search_after_date_filter="1/1/2025",
    search_before_date_filter="12/31/2025",
)
You can also pass any parameter as an AG2 Variable so it can be resolved from the conversation context at runtime: 
from autogen.beta import Variable
from autogen.beta.tools.search import PerplexitySearchToolkit

toolkit = PerplexitySearchToolkit()
search = toolkit.search(
    max_results=Variable("user_max"),
    search_recency_filter=Variable(),  # resolved from the variable named after the parameter
)

​
Response Shape

Each call returns a PerplexitySearchResponse with the original query and a list of PerplexitySearchResult items: 
@dataclass(slots=True)
class PerplexitySearchResult:
    title: str
    url: str
    snippet: str | None = None
    date: str | None = None

@dataclass(slots=True)
class PerplexitySearchResponse:
    query: str
    results: list[PerplexitySearchResult]
    content: str | None = None
    citations: list[str]
    images: list[PerplexityImageMeta]

AG2 Docs

Build multi-agent systems with AG2.

Tool Source

PerplexitySearchToolkit implementation.

Search API Quickstart

Learn more about the underlying Perplexity Search API.

Common Toolkits

Browse all built-in AG2 tools.

​
Support

Need help with the integration?