87 /100
Strong for local analytical tasks when agents are constrained to read-only queries.
Structured output 92/100
Whether the CLI can return JSON, YAML, or stable machine-readable output.
Non-interactive use 88/100
Whether common workflows can run predictably without an interactive prompt.
Safety boundaries 68/100
Whether read-only, dry-run, scoped auth, and destructive command boundaries are clear.
安装方式
Homebrew
$ brew install duckdb pip
$ pip install duckdb 常用命令
Query CSV · 安全
$ duckdb -c "SELECT * FROM 'data.csv' LIMIT 5" Reads local tabular data quickly.
Export JSON · 安全
$ duckdb -json -c "SELECT count(*) AS rows FROM 'data.csv'" Returns parseable JSON output.
Delete rows · 高风险
$ duckdb data.db -c "DELETE FROM events WHERE ..." Mutates data and requires approval.
Agent 使用示例
Gemini CLI
Use DuckDB for read-only analysis of CSV, Parquet, and local databases. Prefer SELECT queries and JSON output. 安全说明
- Local analytical reads are generally safe.
- COPY, INSERT, UPDATE, DELETE, and external database access should be reviewed.
- Agents should run SELECT-only queries by default when exploring data.
Agent 可以用它做什么
DuckDB CLI 可以让 Agent 用 SQL 检查本地数据文件，并输出结构化摘要。
适合场景
适合 CSV 分析、Parquet 检查、本地分析和可重复的数据校验。