Reference · Glossary
Agentic coding
Last updated
Using an AI **agent** — not just autocomplete — to plan, edit multiple files, run terminal commands, and iterate on its own until a task is done, with you reviewing checkpoints instead of typing every line.
#When to use
Multi-file changes, repetitive refactors, writing tests against existing code, or exploring an unfamiliar codebase where the agent can search and summarize faster than you can by hand.
#When not to
Security-sensitive or hard-to-reverse changes (deleting data, touching auth, production infra) without a human review step — agent autonomy should scale with how easy the change is to verify and undo.
#Example
"Add a `/health` endpoint, write a test for it, and run the test suite" — a single instruction an agent can turn into a plan, edits, a shell command, and a pass/fail result, versus writing each file yourself.