75 /100
psql is useful for agents when usage is constrained to predictable commands and documented output.
Installability 79/100
Availability through common package managers and simple install commands.
Structured output 75/100
JSON, YAML, CSV, markdown, or stable machine-readable output for agents.
Safety 61/100
Clear read-only commands, dry-run options, and approval gates for destructive work.
Agent docs 67/100
How easily an agent can learn the command surface from docs, help, and examples.
Install options
Recommended
$ brew install libpq Common commands
Show help · safe
$ libpq --help Inspect command surface before automation.
Check version · safe
$ libpq --version Verify the installed CLI.
Agent usage examples
Claude Code / Codex CLI / Gemini CLI
Use psql only with non-interactive commands. Prefer structured output: csv, text, json. Ask before destructive operations. Safety notes
- Prefer non-interactive usage and explicit output flags when an agent runs this command.
- Treat commands that deploy, delete, charge money, or mutate production data as approval-required.
- Record the exact command, output format, and scope before adding it to an agent workflow.
What agents can use it for
Query PostgreSQL databases and inspect schemas from terminal and agent workflows.
Agent usage rule
Prefer non-interactive commands, structured output, and approval gates for state-changing operations.
Install
brew install libpq