Skip to content

AI agents: choose a place to start

Start with the task and the environment it needs. An agent that can work in a local project may need another connection to reach your notes or company data.

Choose the work before the interface

For a code project, look for file editing, command execution and a way to inspect changes. For research or documents, check the supported inputs and output formats. A terminal, desktop window or browser is an interface choice; it does not establish what the agent can do.

Check how the model is paid for

A tool may use an included account, separately billed API usage or a local model. Check the official requirements for your chosen sign-in method. A subscription to a chat product does not automatically establish access to every API or agent.

Make the first result easy to inspect

Try a small task on copies of files. Ask which inputs were used and inspect the saved output or diff. Choose the permissions for that task before allowing broader access.

Examples from the Library

These pages explain the projects and link to their creators. The selection illustrates different uses; it is not a ranking.

Browse all agents to filter by task, cost and setup effort.

Before you install

Install your first AI agent explains the next checks. If you need help with a specific setup, describe the task and where it stopped.

Source

Anthropic: Building effective agents. Project-specific sources and installation steps are on each Library page.