Crush logo

Crush

Community
NatalieCao323
crush

Build a living AI model of your crush from chat logs, photos, and social media. Runs Bayesian memory tagging, attachment-style diagnosis, and generates a personalized Skill that talks like them. | 把你的 crush 蒸馏成 AI Skill,导入聊天记录,运行贝叶斯信号分析,诊断依恋类型,并模拟对话演习。

Overview

PublisherNatalieCao323
Repositorycrush-skill
Skill namecrush
Stars
61
Forks
2
Bundled files
20
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 20 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by NatalieCao323 on GitHub. Read the source before you install it.

Installation

Install the Crush AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/NatalieCao323/crush-skill/tree/main
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/NatalieCao323/crush-skill.git \
  .claude/skills/crush
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Crush in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Crush on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Crush is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Language: Detect the user's language from their first message and respond in the same language throughout. This skill supports English and Chinese.

crush.skill

Inspired by ex-skill.


Platform Compatibility

Claude Code (claude CLI):

  • All slash commands work natively.
  • Python tools run in your local shell via the Bash tool.
  • ${CLAUDE_SKILL_DIR} resolves to the skill directory automatically.
  • No additional configuration required beyond pip install -r requirements.txt.

OpenClaw:

  • Install to ~/.openclaw/skills/crush or <workspace>/skills/crush.
  • The metadata.openclaw.requires.bins gate ensures the skill loads only when python3 is on PATH.
  • Use {baseDir} in place of ${CLAUDE_SKILL_DIR} — OpenClaw resolves this at runtime.
  • Slash commands are exposed as user-invocable commands via the Skills UI.
  • In sandboxed mode, python3 and Pillow/piexif must also be available inside the container. Install via agents.defaults.sandbox.docker.setupCommand.
  • The metadata.openclaw.install block enables one-click dependency installation from the macOS Skills UI.

Setup

bash
pip install -r requirements.txt

Trigger Conditions

Start the intake flow when the user says any of the following:

  • /create-crush
  • "Help me create a crush skill"
  • "I want to analyze a situationship"
  • "New crush" / "Make a skill for [name]"
  • "I want to practice talking to [name]"

Enter Evolution Mode when:

  • "I found more chat logs" / "Append new data"
  • "They wouldn't say that" / "That's not right"
  • /update-crush {slug}

Tool Usage

TaskTool
Read PDF / imagesRead
Read MD / TXT filesRead
Parse WeChat exportsBash → python3 ${CLAUDE_SKILL_DIR}/tools/wechat_parser.py
Parse QQ exportsBash → python3 ${CLAUDE_SKILL_DIR}/tools/qq_parser.py
Parse social media textBash → python3 ${CLAUDE_SKILL_DIR}/tools/social_parser.py
Analyze photo EXIF / GPSBash → python3 ${CLAUDE_SKILL_DIR}/tools/photo_analyzer.py
Bayesian message taggingBash → python3 ${CLAUDE_SKILL_DIR}/tools/bayesian_tagger.py
Version snapshotsBash → python3 ${CLAUDE_SKILL_DIR}/tools/version_manager.py
Write / update skill filesWrite / Edit
List existing profilesBash → python3 ${CLAUDE_SKILL_DIR}/tools/skill_writer.py --action list

OpenClaw note: Replace ${CLAUDE_SKILL_DIR} with {baseDir} in all Bash commands.

Output directory: ./crushes/{slug}/ relative to the current workspace.


Safety Rules

  1. For personal emotional analysis and conversation practice only. Not for harassment, stalking, or any purpose that violates another person's privacy.
  2. The generated Skill is a simulation. It does not replace real communication and should not be used to deceive others.
  3. If the user shows signs of unhealthy fixation or emotional harm, flag it directly and suggest stepping back.
  4. All data is processed and stored locally. Nothing is uploaded to external servers.
  5. The generated crush Skill will not fabricate statements the real person would never make (e.g., sudden confessions) unless supported by high-confidence evidence in the source material.

Main Workflow: Create a New Crush Profile

Step 1 — Intake

Follow ${CLAUDE_SKILL_DIR}/prompts/intake.md. Ask three questions:

  1. Name or alias (required)
  2. Basic background (optional)
  3. Personality impression (optional)

Step 2 — Import Raw Materials

Ask the user to provide source data. Supported formats:

FormatParser
WeChat TXT / JSON exportwechat_parser.py
QQ TXT exportqq_parser.py
Social media posts / notessocial_parser.py
Photos with EXIF dataphoto_analyzer.py
Direct paste or dictationNo parser needed

Step 3 — Analysis Pipeline

Run in this order:

  1. Parse: Run the appropriate parser(s) on uploaded files.
  2. Bayesian tagging: Run bayesian_tagger.py --file <parsed_output> to tag all messages.
  3. Memory construction: Follow prompts/memory_builder.md to generate memory.md.
  4. Persona construction: Follow prompts/persona_builder.md to generate persona.md.
  5. Merge: Follow prompts/merger.md to combine everything into the final SKILL.md.

Step 4 — Preview and Save

Show the user a summary:

Crush Health Report — [Name]

Bayesian Progression Index: [score]/10
Attachment Style: [type]
Primary Psychological Environment: [type]
Core Issue: [one sentence]
Recommended Strategy: [one sentence]

If the user confirms, write files:

bash
mkdir -p crushes/{slug}
# Write: crushes/{slug}/memory.md
# Write: crushes/{slug}/persona.md
# Write: crushes/{slug}/SKILL.md
python3 ${CLAUDE_SKILL_DIR}/tools/version_manager.py --action save --slug {slug} --message "Initial creation"

Inform the user:

Crush profile created.

Location: crushes/{slug}/
Commands:
  /{slug}          Simulation mode — practice conversations
  /{slug}-report   Advisor mode — full analysis and strategy

Evolution Mode

When the user provides corrections or new data:

  1. Follow ${CLAUDE_SKILL_DIR}/prompts/correction_handler.md.
  2. Update persona.md and/or memory.md as needed.
  3. Re-run merger.md to regenerate SKILL.md.
  4. Save a new version snapshot.

Management Commands

/list-crushes — List all profiles:

bash
python3 ${CLAUDE_SKILL_DIR}/tools/skill_writer.py --action list --base-dir ./crushes

/update-crush {slug} — Append new data to an existing profile.

/versions {slug} — List version history:

bash
python3 ${CLAUDE_SKILL_DIR}/tools/version_manager.py --action list --slug {slug}

/rollback {slug} {version_id} — Restore a previous version:

bash
python3 ${CLAUDE_SKILL_DIR}/tools/version_manager.py --action rollback --slug {slug} --version {version_id}

/delete-crush {slug} — Delete a profile permanently:

bash
rm -rf crushes/{slug}

/wake-up {slug} — Alias for delete. Outputs: "You are awake now."

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Crush AI skill do?

Build a living AI model of your crush from chat logs, photos, and social media. Runs Bayesian memory tagging, attachment-style diagnosis, and generates a personalized Skill that talks like them. | 把你的 crush 蒸馏成 AI Skill,导入聊天记录,运行贝叶斯信号分析,诊断依恋类型,并模拟对话演习。

Why use Crush on TypingMind?

Because you install it once and use it with any model. Crush is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Crush in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/NatalieCao323/crush-skill/tree/main. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Crush?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Crush?

As many as you like. As long as a model supports skills, you can use Crush with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Crush AI skill free?

Yes. It is published on GitHub by NatalieCao323 under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇