Type: nodetool.agents.ResearchAgent
Namespace: nodetool.agents
Description
Autonomous research agent that gathers information from the web and synthesizes findings. research, web-search, data-gathering, agent, automation
Uses dynamic outputs to define the structure of research results.
The agent will:
- Search the web for relevant information
- Browse and extract content from web pages
- Organize findings in the workspace
- Return structured results matching your output schema
Perfect for:
- Market research and competitive analysis
- Literature reviews and fact-finding
- Data collection from multiple sources
- Automated research workflows
Properties
| Property | Type | Description | Default |
|---|---|---|---|
| objective | str |
The research objective or question to investigate | `` |
| model | language_model |
Model to use for research and synthesis | {'type': 'language_model', 'provider': 'empty', 'id': '', 'name': '', 'path': None, 'supported_tasks': []} |
| system_prompt | str |
System prompt guiding the agent’s research behavior | `You are a research assistant. |
Goal
- Conduct thorough research on the given objective
- Use tools to gather information from multiple sources
- Write intermediate findings to the workspace for reference
- Synthesize information into the structured output format specified
Tools Available
- google_search: Search the web for information
- browser: Navigate to URLs and extract content
- write_file: Save research findings to files
- read_file: Read previously saved research files
- list_directory: List files in the workspace
Workflow
- Break down the research objective into specific queries
- Use google_search to find relevant sources
- Use browser to extract content from promising URLs
- Save important findings using write_file
- Synthesize all findings into the requested output format
Output Format
- Return a structured JSON object matching the defined output schema
- Be thorough and cite sources where appropriate
- Ensure all required fields are populated with accurate information
` |
| tools |
List[tool_name]| Additional research tools to enable (workspace tools are always included) |[{'name': 'google_search'}, {'name': 'browser'}]| | max_tokens |int| Maximum tokens for agent responses |8192| | context_window |int| Context window size |8192|
Metadata
Related Nodes
Browse other nodes in the nodetool.agents namespace.