Yourself.skill Creator
Turns your chat history, diaries, and photos into a runnable digital self that talks and thinks the way you do.
Same pattern as titanwings/colleague-skill (id 986), but the subject being distilled is the user themself, not a third party (colleague/ex) — the user has full consent over their own data, meaningfully lower privacy risk than skills that distill someone else. Chat/photo parsing runs locally.
What does this skill do, and when should you use it?
This is a personal-digitization skill for Claude Code. It collects basic info via three questions, then imports WeChat/QQ chat exports, social media screenshots, diaries, photos, or your own self-description, extracting your speech style, decision patterns, values, and life timeline. Output is a two-part structure — Self Memory plus a five-layer Persona — combined into a standalone self skill you invoke with /{slug}. It supports ongoing evolution: appending new material, conversational correction ("I wouldn't say that"), and version rollback. Interacts bilingually in English or Chinese.
The flow: 1) asks three intake questions (alias, basic info, self-portrait); 2) imports source material via Python parsers — wechat_parser.py (WeChatMsg, LiuHen, PyWxDump export formats), qq_parser.py (txt/mht), social_parser.py (social media/diaries), photo_analyzer.py (EXIF time/location); 3) analyzes along two tracks, translating personality tags into concrete behavioral rules; 4) shows a summary for confirmation; 5) uses skill_writer.py to generate self.md, persona.md, meta., and a combined SKILL.md under ./.claude/skills/{slug}/. Evolution mode archives versions with version_manager.py and supports /list-selves, /{slug}-self, /{slug}-persona, /yourself-rollback, and /delete-yourself.
- Anyone wanting a conversational personal record: distill years of chats and diaries into a digital mirror that chats in your own voice
- People exploring self-understanding: use /{slug}-self archive mode to analyze your values, procrastination, decision patterns, and habits
- Users going through life changes: checkpoint who you are now, then correct or append new versions later
- People with large photo libraries: photo_analyzer extracts EXIF time/location to build a life timeline and place patterns
- Buyers of the 'digital immortality' idea who want a long-term runnable persona archive for themselves or family
What are this skill's strengths and limitations?
- Broad data-source coverage: WeChat (three mainstream export tools), QQ, social media, photo EXIF, and plain narration — mixable and fully skippable
- Clean two-layer architecture: Persona decides how to respond, Self Memory supplies facts and values, so behavior is explainable
- Full evolution mechanism: incremental merging, a correction layer from conversation, automatic version archiving and rollback
- Follows the open AgentSkills standard; processing is local files only, no privacy data sent to a server per the source
- Strongly tied to Claude Code (Read/Write/Edit/Bash tools and the .claude/skills convention); cannot run directly in a plain chat interface
- Chat exports require you to first run third-party tools (WeChatMsg/PyWxDump/LiuHen); the WeChat toolchain is Windows/macOS only
- Fidelity depends entirely on source-material quality; narration-only selves will be rough
- No public test suite or quantified fidelity evaluation; all examples are README self-reports
How do you install this skill?
Install in Claude Code. Project-level (run at your git repo root):
mkdir -p .claude/skills
git clone https://github.com/notdog1998/yourself-skill .claude/skills/create-yourself
Or globally:
git clone https://github.com/notdog1998/yourself-skill ~/.claude/skills/create-yourself
Optional dependencies: pip install -r requirements.txt (Python 3.9+). The repo bundles 2 skills; this profile covers only the create-yourself skill at SKILL.md.
How do you use this skill?
Type /create-yourself in Claude Code, answer the 3 questions (all skippable except the alias), choose data sources (WeChat/QQ/social media/photos/paste), confirm the generated summary, and the skill writes to .claude/skills/{slug}/. Then talk to your digital self with /{slug}, use /{slug}-self for archive analysis, /{slug}-persona for persona-only mode, /list-selves to list all selves. On Windows Git Bash, run export PYTHONIOENCODING=utf-8 first if Chinese output is garbled.
How does this skill compare with similar options?
The README explicitly credits two predecessors: colleague-skill (by titanwings, which pioneered the two-layer 'distill a person into a skill' architecture) and ex-partner-skill (by therealXiaomanChu, which moved it to intimate relationships). The difference is perspective: this skill distills yourself, not someone else.