Everything on Technical Risk

LLMs exhibit a bias toward re-implementing functionality rather than reusing existing libraries, leading to potential security and maintenance issues. This bias must be actively countered through explicit context rules that enforce reuse.

Impact: Failing to address this bias results in increased technical debt and security vulnerabilities, as re-implemented code lacks the benefits of centralized updates and patches.

— from Agentic Development: Context as Core Competency · The AI Native Dev - from Copilot today to AI Native Software Development tomorrow· Feb 17, 2026