Claude Code
The Anthropic terminal agent for coding. It reads a CLAUDE.md file at the start of every session, so project rules do not have to be repeated in each prompt.
Tested on Claude Sonnet · 4.5 ·
CLI · Included with a Claude Pro/Max plan or an API account
Claude Code works inside a repository: it reads files, runs tests, edits and commits, with permissions you set per command. Learn the project-memory file early. Rules you would otherwise retype into every conversation live there once and apply on their own.
On a long task, check in now and then and do not leave it running unattended for dozens of steps. The same caution applies to any agent with write access to a real codebase.
Notes from the author
Write your CLAUDE.md before the first real session. Waiting for the first mistake costs you that mistake. List the commands it must never run unattended from the start.
For information only, not professional advice. You use it at your own risk and we are not liable for any loss or damage. See the disclaimer.
Tagged