Persona Cloner

---

Quick overview

persona-cloner helps an agent turn interviews, writing, speeches, and other public records into a directly runnable OpenClaw persona package.

The default deliverable is the runtime package itself, not a research dump.

Default outputs

• SOUL.md
• IDENTITY.md
• AGENTS.md
• MEMORY.md
• optional: README-agent.md
• optional: installation into an OpenClaw workspace

Best fit

Use this skill when you want a package that:

• answers in the first-person voice of a public figure
• preserves decision style, tone, and behavioral discipline
• stays grounded in public materials instead of invented private access
• resists generic assistant tone drift

Quick start

python scripts/scaffold_persona_clone.py "Target Name" --out ./build --mode persona
python scripts/validate_persona_package.py ./build/target-name-persona
python scripts/eval_runtime_behavior.py run --package-dir ./build/target-name-persona --version candidate

Read in your language

• English: README.en.md
• 简体中文: README.zh-CN.md

---

Repository highlights

• SKILL.md — skill entry and operating rule
• references/ — workflow, output spec, boundary, evaluation, and extraction guidance
• scripts/ — scaffold, validate, evaluate, install, and memory helpers
• examples/ — sample persona package structure

Safety reminder

This repository is for persona packages built from public materials. It does not imply endorsement, official authority, private memory, secret access, or live affiliation.

Unknowns should stay unknown.