A curated directory of product CLIs, developer tools, and terminal apps — verified for humans, scripts, and AI agents.
>_$ duckdb
Query local files and analytical datasets with an in-process SQL database.
>_$ neonctl
Manage Neon projects, Postgres branches, databases, roles, connection strings, and API keys.
>_$ supabase
Run Supabase locally, manage database migrations, deploy Edge Functions, and generate database types.
>_$ datasette
Explore and publish structured data through a CLI and web interface.
>_$ hasura
Manage Hasura metadata, migrations, seeds, console sessions, and GraphQL Engine projects.
>_$ atlas
Manage MongoDB Atlas clusters, projects, users, networking, backups, and local deployments.
>_$ mongosh
The official interactive JavaScript shell for MongoDB.
>_$ pscale
Manage PlanetScale databases, branches, deploy requests, passwords, shells, and secure connections.
>_$ prisma
Model databases, generate clients, create migrations, inspect schemas, and browse data.
>_$ turso
Create and operate Turso databases, replicas, groups, tokens, and embedded replica workflows.
>_$ litecli
A SQLite terminal client with completion and formatted tables.
>_$ mycli
A MySQL terminal client with autocompletion and syntax highlighting.
>_$ pgcli
A PostgreSQL terminal client with smart completion and highlighting.
>_$ sqlite-utils
Create, transform, query, and inspect SQLite databases from scripts.
>_$ usql
A universal command-line interface for SQL and NoSQL databases.
Common questions about CLI tools, Agent CLIs, and command line automation
A CLI, or command-line interface, lets you control software by typing commands in a terminal instead of clicking through a graphical interface. CLI tools are useful for development, automation, scripts, CI pipelines, and repeatable operations.
An Agent CLI is either a command-line tool for operating an AI agent or a conventional CLI designed to be called reliably by AI agents. Agent-friendly CLIs expose clear commands, non-interactive options, predictable output, and meaningful exit codes so an agent can plan, execute, and verify work.
Yes. AI coding agents can invoke many of these CLIs through a shell to work with repositories, cloud services, databases, productivity apps, and deployment systems. Tools marked Agent-ready are especially suitable for scripted or autonomous use, but authentication and permissions still need to be configured safely.
The best CLI tools for AI agents support non-interactive commands, stable flags, structured or parseable output, standard input and output, documented authentication, and reliable exit codes. These features make results easier for an agent to interpret and verify without manual intervention.
Choose an official CLI when you need first-party support, close alignment with a service API, and predictable updates. A community CLI can be a better fit when it offers a simpler workflow or capabilities the official tool lacks. Review its repository, maintenance activity, permissions, and installation instructions before use.
Use an OS keychain, environment variables, or a dedicated secret store instead of placing API keys in repositories, prompts, or shell history. Give people and AI agents least-privilege access, start in read-only or test environments, and confirm that automated commands support previews, explicit confirmation controls, and reliable exit codes.
