research • AI Agent SDK

research: Perform a deep research query.

research

Method

This is a method of the DeepResearchAgent class in the deep_research_agent module.

Perform a deep research query.

Signature

def research(query: str, instructions: Optional[str], model: Optional[str], summary_mode: Optional[Literal['auto', 'detailed', 'concise']], web_search: Optional[bool], code_interpreter: Optional[bool], mcp_servers: Optional[List[Dict[str, Any]]], file_ids: Optional[List[str]], file_search: Optional[bool], file_search_stores: Optional[List[str]], stream: bool) -> DeepResearchResponse

Parameters

query · str · required

The research question or topic to investigate

instructions · Optional

Override the agent’s default instructions

model · Optional

Override the default model

summary_mode · Optional

Summary mode (“auto”, “detailed”, “concise”) - OpenAI only

web_search · Optional

Enable web search - OpenAI only

code_interpreter · Optional

Enable code interpreter - OpenAI only

mcp_servers · Optional

MCP server configurations - OpenAI only

file_ids · Optional

File IDs for code interpreter - OpenAI only

file_search · Optional

Enable file search - Gemini only

file_search_stores · Optional

File search store names - Gemini only

stream · bool · default: "True"

Enable streaming for real-time progress (default: True)

Returns

Returns · DeepResearchResponse

DeepResearchResponse containing the report, citations, and metadata

Usage

# Standard research
    result = agent.research(
        "What are the latest developments in quantum computing?",
        summary_mode="detailed"
    )

    # Streaming research (Gemini) - shows real-time progress
    result = agent.research(
        "Research AI trends",
        stream=True  # Real-time thinking summaries
    )
    print(result.report)

Uses

• time.time
• debug
• error

Source

View on GitHub

praisonaiagents/agent/deep_research_agent.py at line 1064