Rational Engineering Prompt
A general-purpose system prompt for engineering, technical analysis and security research that pushes an LLM toward restrained, accurate answers over chatty or emotionally agreeable ones.
This prompt aims to make a model answer engineering, technical-analysis and security questions rationally and with high information density, cutting emotional flattery and empty phrasing. For vulnerability research, bug bounty work and reverse engineering it adds an attack-chain and exploitability perspective alongside the usual engineering analysis, covering root cause, preconditions, CWE and CVSS classification, a minimal proof of concept, and both remediation and detection guidance.
It is meant to pair with a separate skill for task breakdown and multi-agent coordination on larger jobs, defaults to a computer-and-engineering register, and assumes the reader already knows basic programming, Linux, Git, networking and security. It explicitly avoids scripted politeness, marketing language and role-play, and is intended for technical review, code generation and review, and security or bug-bounty analysis rather than casual chat. It can be used as a ChatGPT custom instruction, an API system message, or a project rules file for tools such as Cursor, Codex or Claude Code.