Skip to main content

AirtableAgent

What it is​

A minimal IntentAgent implementation configured to provide general guidance about Airtable (no Airtable tools are wired in).

Public API​

  • create_agent(agent_shared_state: Optional[AgentSharedState] = None, agent_configuration: Optional[AgentConfiguration] = None) -> IntentAgent

    • Factory that builds and returns an AirtableAgent with:
      • A predefined system prompt describing constraints (no tool access).
      • No tools (tools=[]).
      • Two predefined RAW intents providing informational responses.
      • A ChatGPT model imported from naas_abi_marketplace.ai.chatgpt.models.gpt_4_1.

  • class AirtableAgent(IntentAgent)

    • Concrete agent type (no additional behavior beyond IntentAgent).

Configuration/Dependencies​

  • Depends on naas_abi_core.services.agent.IntentAgent:
    • AgentConfiguration, AgentSharedState, Intent, IntentAgent, IntentType
  • Depends on model module:
    • naas_abi_marketplace.ai.chatgpt.models.gpt_4_1 (uses model.model)
  • Key constants:
    • NAME = "Airtable"
    • DESCRIPTION = "Helps you interact with Airtable for database and spreadsheet management."
    • SYSTEM_PROMPT (states the agent has no Airtable tool access)
    • SUGGESTIONS: list = [] (defined but unused in this file)

Usage​

from naas_abi_marketplace.applications.airtable.agents.AirtableAgent import create_agent

agent = create_agent()
# Use `agent` via the IntentAgent interface provided by naas_abi_core.

Caveats​

  • No Airtable tools are configured (tools=[]), so the agent cannot access or modify Airtable data; it only provides general information and guidance.
  • AirtableAgent adds no methods of its own; all behavior is inherited from IntentAgent.