immortal-skill

Install

npx skillfoxx add skills/immortal-skill

npx skills add agenmod/immortal-skill --skill immortal-skill -a claude-code -y

npx skills add agenmod/immortal-skill --skill immortal-skill -a cursor -y

npx skills add agenmod/immortal-skill --skill immortal-skill -a github-copilot -y

npx skills add agenmod/immortal-skill --skill immortal-skill -a codex -y

npx skills add agenmod/immortal-skill --skill immortal-skill -a gemini-cli -y

npx skills add agenmod/immortal-skill --skill immortal-skill -a cline -y

npx skills add agenmod/immortal-skill --skill immortal-skill -a roo -y

npx skills add agenmod/immortal-skill --skill immortal-skill -a opencode -y

git clone https://github.com/agenmod/immortal-skill.git
cp -r immortal-skill ~/.openclaw/skills/immortal-skill

This is third-party code. Review the repository files before installing.

What it does

The skill turns chat logs, documents and messenger exports into a structured profile of a person that an agent can load as a persona. It works across four dimensions: how the person acts, how they communicate, what they remember and what kind of person they are; each fact is tagged with an evidence level from verbatim quote to impression, and contradictions are kept rather than smoothed over. Seven role templates cover the user themselves, a colleague, a mentor, a family member, a partner, a friend and a public figure, each with its own consent and anonymization requirements. Data collection spans twelve-plus sources via API or local database reads, including WeChat, Feishu, iMessage, Telegram, Slack, Discord, email and manual file import. The output is a folder with a SKILL.md and dimension files compatible with the Agent Skills standard and OpenClaw. The repository also bundles related modules: distilling public figures' frameworks from open material, defending your own data against unwanted collection and a permissions protocol for persona use.

Who it is for. For people who want to build a digital twin of themselves or, with consent, of relatives, mentors and colleagues for use with an agent.

Example request

Build a digital twin from my Telegram and Twitter chats and package it as an Agent Skill

Limitations

The skill works with private correspondence, so applying it to other people's data is only acceptable with their consent and with anonymization, and that responsibility is on the user. Some collectors need Python 3.9+, platform API tokens or access to local messenger databases, and availability of specific platforms from your network may vary. Profile quality depends on the volume and cleanliness of source data and on the agent model that performs the extraction. The prebuilt public-figure personas in the repository are assembled from open material and do not imply those people's stance or authorization.

How to disable. Remove the immortal-skill folder from your agent's skills directory, for example ~/.openclaw/skills/ or the project's .claude/skills/. Generated profiles live in separate folders and are removed manually.

Security check

• Reads private correspondence and local messenger databases, handling sensitive data
• Uses platform API tokens and runs Python collectors that access the network

README in short

The README presents an open digital twin framework: from chat, social media and documents across twelve-plus sources it extracts a person's profile along four dimensions with evidence levels and preserved contradictions. Seven role templates carry different consent and anonymization requirements, and it runs through conversation with the agent or through the immortal_cli.py CLI. The output is packaged as a folder with a standard SKILL.md compatible with Agent Skills and OpenClaw. The repository bundles related modules: distilling public figures' frameworks from open material, three-layer defense of your own data and a persona permissions protocol. The project stresses ethics and ships example profiles, MIT licensed.

SKILL.md

---
name: immortal-skill
description: "Universal digital immortality framework: distill anyone's digital twin from chat logs, social media and documents across platforms, with seven role templates and 12+ data sources."
license: MIT
metadata: {"openclaw": {"requires": {"bins": ["python3"]}}, "personas": ["self", "colleague", "mentor", "family", "partner", "friend", "public-figure"], "platforms": ["feishu", "dingtalk", "wechat", "imessage", "telegram", "whatsapp", "slack", "discord", "email", "twitter", "social-archive", "manual"]}
---

# Digital immortality

## When to activate

- The user asks to distill someone, build a digital twin, preserve their way or memory, or make the AI act like a person.
- The user provides material about a person and wants a loadable Agent Skill package.

## Core idea

Choose a role, collect across platforms, extract by dimension (procedure, interaction, memory, personality), grade evidence, merge conflicts, output a skill directory compatible with Agent Skills.

Related