Create Agent¶
Builds an agent configuration by selecting a language model, enabling external tools, and defining the agent’s role and system prompt. It does not run any inference; it prepares a reusable AGENT object for downstream nodes. Model and tool options are dynamically fetched from the Agents service with sensible fallbacks.
Usage¶
Use this node at the start of an agents workflow to define a single agent’s capabilities and behavior. Connect its AGENT output into nodes that orchestrate or run agents (e.g., a team runner) to execute tasks with the chosen model and tools.
Inputs¶
| Field | Required | Type | Description | Example |
|---|---|---|---|---|
| model | True | SELECT | Select the LLM to power the agent. Options are fetched from the Agents service; if unavailable, a fallback list is used. | gpt-4o |
| tools | True | MULTICOMBO | Choose external tools the agent may call during reasoning. Each option can be toggled on/off. | {'Calculator': True, 'DateTime': False, 'WebScraper': False, 'WebSearcher': True, 'JSONExtractor': False, 'CSVExtractor': False} |
| role | True | STRING | Short role description used to characterize the agent. This text is prepended to the system prompt and may be used when selecting agents in a team. | A helpful research assistant that finds and summarizes reputable sources. |
| system_prompt | True | STRING | The agent’s core instructions: personality, scope, constraints, and behavior guidelines. | You are precise and cite sources. Prefer concise answers with bullet points when appropriate. |
Outputs¶
| Field | Type | Description | Example |
|---|---|---|---|
| agent | AGENT | An agent configuration object containing the selected model, tools map, role, and system prompt. Use this as input to nodes that run or orchestrate agents. | {'model': 'gpt-4o', 'tools': {'Calculator': True, 'DateTime': False, 'WebScraper': False, 'WebSearcher': True, 'JSONExtractor': False, 'CSVExtractor': False}, 'role': 'A helpful research assistant that finds and summarizes reputable sources.', 'system_prompt': 'You are precise and cite sources. Prefer concise answers with bullet points when appropriate.'} |
Important Notes¶
- Dynamic options: Model and tool lists are fetched from the Agents service at runtime; if the service is unavailable, a fallback list is shown.
- No execution: This node only constructs an AGENT config; it does not generate messages or call the model.
- Role usage: The role is prepended to the system prompt and may influence agent selection when used in teams.
- Tools control capabilities: If no tools are selected, the agent relies solely on the LLM with no external capabilities (e.g., browsing, scraping, or parsing).
- Compatibility: The AGENT output is intended for use with agents orchestration nodes (e.g., team runners).
Troubleshooting¶
- Empty or missing model/tool options: The Agents service may be unreachable. Retry, check network access, or use the fallback options presented.
- Unexpected model unavailability at runtime: Ensure the selected model is supported by your environment and service configuration.
- Agent lacks expected capabilities: Verify the corresponding tools are enabled in the tools multiselect.
- Downstream node errors: If a runner node fails using this agent, confirm the AGENT output is connected correctly and adjust any runtime limits (e.g., timeouts) in the runner node.