Le Chat
  • Overview
Mistral Vibe
Studio

Publish in le Chat

To publish a conversational workflow as an assistant in le Chat (Mistral's chat interface), your workflow must return a ChatAssistantWorkflowOutput. The output won't be displayed in le Chat, but we enforce a common interface for inter-operability.

import mistralai.workflows as workflows
import mistralai.workflows.plugins.mistralai as workflows_mistralai


@workflows.workflow.define(
    name="expense-summary-workflow",
    workflow_display_name="Expense Summary",
    workflow_description="Generates a summary of expenses",
)
class ExpenseSummaryWorkflow:
    @workflows.workflow.entrypoint
    async def run(self, department: str) -> workflows_mistralai.ChatAssistantWorkflowOutput:
        summary = f"Expense summary for {department}: Total $12,500"

        return workflows_mistralai.ChatAssistantWorkflowOutput(
            content=[workflows_mistralai.TextOutput(text=summary)]
        )
Tagging input variants

Tagging input variants

When a workflow accepts a union of input types, clients may need a way to know which variant to use. The @input_tag decorator adds an x-input-tag field to a model's JSON schema so clients can identify and select the right one.

from pydantic import BaseModel

import mistralai.workflows as workflows
import mistralai.workflows.plugins.mistralai as workflows_mistralai
from mistralai.workflows.plugins.mistralai import input_tag


class FullParams(BaseModel):
    config: str
    options: dict[str, str]


@input_tag("simplified")
class SimpleParams(BaseModel):
    prompt: str


@workflows.workflow.define(
    name="multi-client-workflow",
    workflow_display_name="Multi-Client",
)
class MultiClientWorkflow:
    @workflows.workflow.entrypoint
    async def run(self, params: FullParams | SimpleParams) -> workflows_mistralai.ChatAssistantWorkflowOutput:
        if isinstance(params, SimpleParams):
            resolved = params.prompt
        else:
            resolved = params.config

        return workflows_mistralai.ChatAssistantWorkflowOutput(
            content=[workflows_mistralai.TextOutput(text=f"Received: {resolved}")]
        )

SimpleParams.model_json_schema() now contains "x-input-tag": "simplified". Clients inspect the union members in the workflow's input_schema and select the variant whose tag they recognise.

Error handling

Error handling

To signal that a workflow has failed, set isError=True on the output. The text content is used as the error message displayed to the user, and the workflow is marked as failed in the conversation:

import mistralai.workflows as workflows
import mistralai.workflows.plugins.mistralai as workflows_mistralai


@workflows.workflow.define(
    name="expense-validation-workflow",
    workflow_display_name="Expense Validation",
    workflow_description="Validates expense data",
)
class ExpenseValidationWorkflow:
    @workflows.workflow.entrypoint
    async def run(self, expense_id: str) -> workflows_mistralai.ChatAssistantWorkflowOutput:
        expense = lookup_expense(expense_id)
        if expense is None:
            return workflows_mistralai.ChatAssistantWorkflowOutput(
                content=[workflows_mistralai.TextOutput(text=f"Expense {expense_id} not found.")],
                isError=True,
            )

        return workflows_mistralai.ChatAssistantWorkflowOutput(
            content=[workflows_mistralai.TextOutput(text=f"Expense {expense_id} is valid.")]
        )
Structured content

Structured content

ChatAssistantWorkflowOutput accepts an optional structuredContent field (dict[str, Any]) for attaching arbitrary structured data to the workflow output.

return workflows_mistralai.ChatAssistantWorkflowOutput(
    content=[workflows_mistralai.TextOutput(text="Done.")],
    structuredContent={"tool": "web_search", "results": [{"url": "https://example.com"}]},
)