bonnard-cli

概述

Bonnard 是一个开源的智能体模式CLI，专为Claude、Gemini、Codex和Copilot优化。它作为语义层连接数据仓库和各种表面，包括通过MCP的AI智能体、Markdown格式的仪表板以及通过REST API的内部工具。该平台确保所有消费者之间的一致指标定义，解决了不同工具间指标漂移的常见问题。它支持Snowflake、BigQuery、Databricks和PostgreSQL等主要数据仓库，并能自动检测现有模式和关系。

试试问 AI

装完之后，这里有 5 个你可以让 AI 做的事：

• bon_init

  Scaffold a new project with agent configs

  试试:Initialize a new Bonnard project试试:Set up a new semantic layer project
• bon_datasource_add

  Connect a data source or add demo data

  试试:Connect to Snowflake data warehouse试试:Add the retail demo dataset
• bon_validate

  Validate YAML syntax locally

  试试:Validate my model definitions试试:Check for YAML syntax errors
• bon_deploy

  Deploy to production with a version message

  试试:Deploy with message 'Q2 metrics update'试试:Push my models to production
• bon_pull

  Download deployed models to local project

  试试:Pull latest production models试试:Update local models with deployed versions
• bon_query

  Run queries (JSON or SQL) against the semantic layer

  试试:Query total revenue by product category试试:Get customer count from orders
• bon_schema

  Explore deployed measures, dimensions, and views

  试试:Show all available metrics试试:List dimensions for the orders view
• bon_dashboard_dev

  Preview a markdown dashboard locally

  试试:Preview my sales dashboard试试:Test the dashboard before deployment
• bon_dashboard_deploy

  Deploy a markdown dashboard

  试试:Deploy my sales dashboard试试:Publish dashboard to production
• bon_mcp

  MCP server setup instructions

  试试:Set up MCP server试试:Configure Bonnard for agent integration
• bon_keys_create

  Create a publishable or secret API key

  试试:Create an API key for my application试试:Generate secret key for SDK access
• bon_docs

  Browse documentation from the CLI

  试试:Show documentation for modeling试试:Get CLI reference guide

可对比工具

安装

npm install -g @bonnard/cli

bon mcp

{
  "mcpServers": {
    "bonnard": {
      "command": "npx",
      "args": ["@bonnard/cli", "mcp"]
    }
  }
}

FAQ

Bonnard支持哪些数据源？

Bonnard支持主要数据仓库，包括Snowflake、Google BigQuery、Databricks、PostgreSQL、Amazon Redshift和DuckDB。它还与dbt、Dagster、Prefect、Airflow和各种BI工具集成。

Bonnard如何确保指标一致性？

Bonnard提供统一的语义层，所有消费者（AI智能体、仪表板、SDK）都查询同一个模式中定义的受控指标，消除了不同工具之间的指标漂移。

bonnard-cli 对比