okhp3-custom-gpt-readiness

原文涵盖的主题

• okhp3-custom-gpt-readiness
• Scope
• Operating contract
• Required output
• Quality gates
• Handoff
• About

OverKill Hill P³ · overkillhill.com · github.com/OKHP3 Turn scattered GPT ideas and partial artifacts into a buildable, testable brief. This is an intake and gap-analysis skill, not the builder itself. | In scope | Out of scope | |---|---| | Concept triage, evidence inventory, requirement elicitation, readiness scoring, and builder handoff | Writing production integrations, managing credentials, publishing a GPT, or claiming volatile platform limits | 1. Inventory before asking. Inspect the supplied conversation, notes, files, prior prompts, prompt-chain stages, and existing GPT export. Record each artifact as present, partial, missing, or conflicting; never ask for information already present. 2. Classify every claim. Quote or point to the source artifact for every present claim. Label it verifiedplatformfact, sourcederivedpractice, theory, preference, inferred, or unknown. A source assertion does not become a platform fact by being repeated. 3. Assess the eight readiness domains: job and audience, outcomes, boundaries, conversation contract, instruction behavior, knowledge/data, tools and permissions, and evaluation/governance. 4. Identify blockers. A missing primary job, audienc

这里只展示节选，请以完整原文和当前要求为准。