MistralAgent
What it is
A thin wrapper around naas_abi_core’s IntentAgent configured to use the Mistral Large model with a predefined system prompt and a small set of coding/documentation-related intents.
Public API
-
Constants
AVATAR_URL: Public avatar image URL.NAME:"Mistral".DESCRIPTION: Short description of the agent/model.SYSTEM_PROMPT: System prompt defining behavior and “self-recognition” rules.SUGGESTIONS: Empty list (no suggestions provided).
-
Functions
create_agent(agent_shared_state: Optional[AgentSharedState] = None, agent_configuration: Optional[AgentConfiguration] = None) -> IntentAgent- Creates and returns a configured
MistralAgent. - Defaults:
AgentConfiguration(system_prompt=SYSTEM_PROMPT)if not provided.AgentSharedState(thread_id="0")if not provided.
- Registers intents that target
call_modelfor:- generate code, review code, optimize code, document technical details, help with programming.
- Creates and returns a configured
-
Classes
class MistralAgent(IntentAgent)- No additional methods/overrides; inherits behavior from
IntentAgent.
- No additional methods/overrides; inherits behavior from
Configuration/Dependencies
- Depends on
naas_abi_core.services.agent.IntentAgent:AgentConfiguration,AgentSharedState,Intent,IntentAgent,IntentType.
- Model import
- Imports
modelfromnaas_abi_marketplace.ai.mistral.models.mistral_large_2411and passesmodel.modelaschat_model.
- Imports
- Tools/Agents
toolsandagentsare empty lists in this implementation.
- Memory
memory=Noneis passed to the agent constructor.
Usage
from naas_abi_marketplace.ai.mistral.agents.MistralAgent import create_agent
agent = create_agent()
# Use `agent` via the IntentAgent interface provided by naas_abi_core.
Caveats
MistralAgentadds no custom logic; all runtime behavior comes fromIntentAgentand the imported chat model.- The registered intents all target
"call_model"; this relies on the underlyingIntentAgentimplementation to route that target correctly.