Skip to main content

QwenAgent

What it is​

A thin wrapper that constructs an IntentAgent configured for the local Qwen3 8B chat model (via Ollama) with a predefined system prompt and a set of intents focused on coding, privacy/local usage, multilingual help, and reasoning.

Public API​

  • Constants

    • NAME: Agent display name ("Qwen").
    • DESCRIPTION: Human-readable description of the agent.
    • AVATAR_URL: URL to an avatar image.
    • SYSTEM_PROMPT: Default system prompt describing capabilities and response style.
    • SUGGESTIONS: Empty list (no default suggestions).

  • Functions

    • create_agent(agent_shared_state: Optional[AgentSharedState] = None, agent_configuration: Optional[AgentConfiguration] = None) -> IntentAgent
      • Creates and returns a configured QwenAgent.
      • Loads the chat model from naas_abi_marketplace.ai.qwen.models.qwen3_8b.
      • Defines a list of Intent objects targeting "call_model".
      • If not provided:
        • Creates AgentConfiguration(system_prompt=SYSTEM_PROMPT)
        • Creates AgentSharedState(thread_id="0")

  • Classes

    • class QwenAgent(IntentAgent)
      • No additional behavior beyond IntentAgent (empty subclass).

Configuration/Dependencies​

  • Depends on naas_abi_core.services.agent.IntentAgent:
    • IntentAgent, Intent, IntentType, AgentConfiguration, AgentSharedState
  • Loads the model dynamically inside create_agent:
    • from naas_abi_marketplace.ai.qwen.models.qwen3_8b import model
  • Tools and sub-agents are currently empty lists (tools = [], agents = []).

Usage​

from naas_abi_marketplace.ai.qwen.agents.QwenAgent import create_agent

agent = create_agent()

# The returned object is an IntentAgent (specifically QwenAgent) configured with:
# - system prompt (SYSTEM_PROMPT)
# - chat model (qwen3_8b model)
# - a predefined set of intents targeting "call_model"
print(type(agent).__name__)

Caveats​

  • QwenAgent itself adds no methods; all behavior comes from IntentAgent.
  • The function assumes naas_abi_marketplace.ai.qwen.models.qwen3_8b.model is importable and correctly configured.