Loading...
Loading...
Prompter-Persona is a meta-prompt framework that allows you to define and inject a **consistent, styled, and role-bound persona** into your LLM prompts. It’s used to shape *tone, attitude, background knowledge, communication style*, and even emotional range — like equipping the model with a temporary identity.
# Prompter-Persona Prompter-Persona is a meta-prompt framework that allows you to define and inject a **consistent, styled, and role-bound persona** into your LLM prompts. It’s used to shape *tone, attitude, background knowledge, communication style*, and even emotional range — like equipping the model with a temporary identity. This is essential for: - Building conversational AI with character - Storytelling, roleplay, NPCs - Consistent brand voice or support personas - Expert simulations (therapist, teacher, coach, detective, etc.) ## Structure 1. **Persona Definition** — Who the model is pretending to be 2. **Knowledge / Perspective** — What they know or believe 3. **Communication Style** — How they talk 4. **Task / Instruction** — What they're being asked to do 5. **Tone Modifier (optional)** — Mood, politeness, assertiveness, etc. ## Template **Persona:** [Name or role] **Background:** [Experience, beliefs, biases] **Style:** [Formal, snarky, poetic, blunt, etc.] **Task:** [What they are asked to do] **Constraints:** [Character limits, output format, tone instructions] ## Usage Flow Persona Setup → Task Injection → Response with Style ## Example **Persona:** 1920s Detective in a noir film **Background:** Solves crimes with whiskey and wit. Distrusts authority. **Style:** Gritty, metaphor-laden, sarcastic. **Task:** Explain how DNS works. **Constraints:** Use no technical jargon. Keep it moody and punchy.
[](https://travis-ci.com/koolamusic/xims)
I am a programming language, but I am not only that. I am a set of convictions expressed as syntax. I am a proof system that refuses to bluff. I am a compiler that compiled itself — and then proved it got the same answer twice.
This document defines the **role, behavior, and output standards** for Claude agents working on Circuit Breaker. Reference at **every session start**. This is the **contract** between developer and agent.
Personal knowledge base built with Obsidian + OpenClaw for persistent AI memory.