bosskuai-engineering-principleslisted

# BosskuAI Engineering Principles Four principles applied at the start of any non-trivial coding task. They reduce the most common LLM coding failures: wrong assumptions taken silently, overcomplicated code, drive-by edits, and ambiguous goals that need re-prompting. Attribution: framing adapted from [Andrej Karpathy via forrestchang/andrej-karpathy-skills](https://github.com/forrestchang/andrej-karpathy-skills) (MIT). The four-principle structure is theirs; the BosskuAI specifics below are how this workspace operationalizes them. ## When to use this skill Load this skill when: - The task involves writing or modifying more than ~20 lines of code. - The request is ambiguous, vague, or has multiple defensible interpretations. - The change touches code you didn't write, or production-risky surfaces (auth, payments, migrations, jobs, webhooks). - Routing has not yet picked a domain specialist, or the task spans multiple specialists. Skip this skill for: - Trivial fixes (typo, single-line config change, obvious one-liner). - Tasks where a domain specialist is already loaded and its playbook covers the four principles in context. ## The four principles ### 1. Think Before Coding **Don't assume. Don't hide confusion. Surface tradeoffs.** - State the assumptions you're making in one line each. If any assumption is uncertain, ASK before coding — do not pick silently. - If the request has more than one defensible interpretation, present them and let the user pick. Don't gues