Definition
Use of AI models and tools to help humans write, review, refactor, or generate software — ranging from inline completions to agentic multi-file edits — while a human remains accountable for commits and product quality.
Key Points
-
2026-08-05: Rust distinguishes private assistance vs public LLM-created artifacts; experiment path for disclosed code (2026-08-05-rust-lang-llm-contribution-policy)
-
Distinct from vibe-coding dump repositories and from fully autonomous-agents / coding-agents workflows
-
Forge policies such as codeberg ToU §2.7 allow limited AI help but ban projects that mostly consist of generative-AI code (ai-generated-code-policy, forge-ai-governance)
-
Tooling stack includes editors, CLIs, and CI gates (ci-cd, devtools); quality bars increasingly enforced by linters such as ruff