Coding agents
What is Vibe Coding?
Vibe coding is directing software work in natural language while agents and tools change the repository.
It still needs requirements, tests, and review. It is not a replacement for verification.
Visual
Vibe coding pipeline
Natural language intent flows through agent planning, editing, testing, and review.
Why it matters
Vibe coding is directing software development in natural language while AI agents handle the actual code changes. You describe what you want; the agent plans, edits, tests, and submits.
Vibe coding does not eliminate the need for requirements, tests, and review. It changes who types the code, not whether the code needs to be correct.
The vibe coding pipeline
A typical vibe coding session follows a pipeline: describe the intent → agent generates a plan → agent edits files → agent runs tests → human reviews the diff → iterate.
Each step produces an observable artifact. The plan is reviewable before execution. The diff is reviewable before merge. Tests provide evidence of correctness.
Key takeaways
- 1Vibe coding changes who types — not whether code needs to be correct.
- 2Tests and review are still required.
- 3The pipeline: intent → plan → edit → test → review → merge.
Common mistakes
- ✕Accepting agent-generated code without review or tests.
- ✕Treating vibe coding as a replacement for understanding the codebase.
- ✕Not maintaining tests that verify agent-generated changes.
Related terms
Concept neighborhood
Terms linked from Vibe Coding in the glossary graph.
- Vibe Coding
- Agentic Coding
- AI Coding Agent