Coding agents
What is AI Coding Agent?
An AI coding agent edits software with repository and test tools in a loop.
Autocomplete predicts tokens. A coding agent plans, patches, and checks.
Why it matters
An AI coding agent uses repository and test tools in a loop to edit software. It plans changes, writes code, runs tests, and iterates. This is fundamentally different from autocomplete, which predicts the next token.
Coding agents make software development faster but do not make it automatic. They still produce bugs, miss edge cases, and need review.
Key takeaways
- 1Coding agents plan, edit, test, and iterate — autocomplete just predicts.
- 2They need repository access, test runners, and review gates.
- 3Agent-generated code needs the same review as human-written code.
Common mistakes
- ✕Merging agent-generated code without review.
- ✕Confusing autocomplete with a coding agent.
- ✕Not running tests after agent edits.
Related terms
Concept neighborhood
Terms linked from AI Coding Agent in the glossary graph.
- AI Coding Agent
- Agentic Coding
- Vibe Coding