Clean Vibe Code Glossary
The definitive vocabulary for AI-era software development.
Core Concepts
Clean Vibe CodingThe methodology for using AI productively while maintaining code ownership, understanding, and sustainability. Emphasizes intentional use over blind delegation.Comprehension DebtThe gap between what code does and what the team understands about it — you own working code you cannot maintain.Synthetic DebtTechnical debt arising from AI-assisted development, characterized by invisibility during accumulation and exponential growth.The Great BifurcationResearch documenting the structural transformation of the software labor market where junior roles are hollowed out while senior/architect roles become more critical.Vibe CodingThe practice of describing functionality to AI in natural language and accepting output based on immediate functionality rather than deep understanding or structural soundness.
Methodology & Management
ADRA formal document that captures a significant architectural decision made during development, including its rationale and trade-offs.Agent BoundariesExplicit, enforceable constraints on an AI agent's scope of action, including allowed/forbidden directories, operations, and decision-making authority.CheckpointA mandatory pause in an autonomous AI agent's execution where a human must review and approve the proposed plan or modifications.Clean PromptA structured prompt pattern that includes explicit context, architectural design decisions, specific tasks, constraints, and required deliverables..cursorrulesA project-level configuration file that provides persistent instructions, constraints, and context to AI coding assistants and agents.Debt BudgetA quantitative limit on the amount of new technical debt a team is permitted to accumulate during a specific development cycle.Debt LedgerA systematic log for tracking technical debt items, including their category, severity, rationale, and planned remediation date.Debt UnitA quantified unit of measure used to categorize and track the severity of technical debt items in a codebase.Human-in-the-Loop (HITL)A development workflow where AI tools require explicit human approval, review, or intervention at critical decision points.IP SegregationA strategic engineering practice of isolating AI-generated 'scaffolding' from human-authored 'Core IP' to maintain legal ownership and copyright protection.Supervised ExecutionAn operational mode for AI agents where autonomy is constrained by mandatory human review and approval at every significant step.
Roles, Risks & Industry
Automation BiasThe psychological tendency to over-trust automated systems (like AI coding assistants), leading to reduced scrutiny and increased confidence in flawed outputs.Bus FactorThe minimum number of team members who have to leave or be 'hit by a bus' before a project comes to a halt due to a lack of knowledge or capability.Digital ArchaeologistA specialized engineering role focused on deciphering, documenting, and reconstructing the design rationale of opaque or poorly understood AI-generated codebases.Technical Due DiligenceThe comprehensive audit of a company's technology, codebase, architecture, and team, usually performed by investors or acquirers to assess technical risk.Over-DelegationA high-severity vibe-code smell occurring when developers grant excessive autonomy to AI assistants or agents, abdicating engineering judgment and oversight.Vibe-Code SmellA specific anti-pattern or characteristic in code or development process that indicates the accumulation of AI-related technical debt.