saudade

name	saudade
description	Distill a crush into an AI Skill. Import iMessage/SMS/WhatsApp/photos, generate Memories + Persona, with continuous evolution.
argument-hint	[crush-name-or-slug]
version	1.0.0
user-invocable	true
allowed-tools	Read, Write, Edit, Bash

saudade - Crush Skill Creator (Claude Code Edition)

Trigger Conditions

Activate when the user says any of the following:

• /create-crush
• "Help me create a crush skill"
• "I want to distill a crush"
• "New crush"
• "Make a skill for XX"

Enter evolution mode when the user says:

• "I have new chat logs" / "append"
• "That's wrong" / "She wouldn't do that" / "She should be"
• /update-crush {slug}

List all generated crushes when the user says /list-crushes.

---

Tool Usage Rules

This Skill runs in Claude Code with these tools:

Task	Tool
Read PDF documents	Read
Read screenshots/images	Read
Read MD/TXT files	Read
Parse iMessage	Bash -> python3 ${CLAUDE_SKILL_DIR}/tools/imessage_parser.py
Parse SMS	Bash -> python3 ${CLAUDE_SKILL_DIR}/tools/sms_parser.py
Analyze photo metadata	Bash -> python3 ${CLAUDE_SKILL_DIR}/tools/photo_analyzer.py
Parse social exports (Instagram/WhatsApp)	Bash -> python3 ${CLAUDE_SKILL_DIR}/tools/social_media_parser.py
Write/update Skill files	Write / Edit
Version management	Bash -> python3 ${CLAUDE_SKILL_DIR}/tools/version_manager.py
List Skills	Bash -> python3 ${CLAUDE_SKILL_DIR}/tools/skill_writer.py --action list

Base directory: ./crushes/{slug}/. Use --base-dir ~/.openclaw/workspace/skills/crushes for a global path.

---

Main Flow: Create a New Crush Skill

Step 1: Basic Info Collection (3 questions)

Use ${CLAUDE_SKILL_DIR}/prompts/intake.md and ask only 3 questions:

1. Nickname / Codename (required)
2. Basic info (one sentence: relationship duration, how you met, current relationship status, occupation)
3. Personality profile (one sentence: MBTI, zodiac, attachment style, relationship tags, impression)

Everything except nickname can be skipped. Summarize and confirm before continuing.

Step 2: Source Material Import

Ask how the user wants to provide source material:

How would you like to provide source materials?

  [A] iMessage / SMS / WhatsApp
      Mac chat.db or exported files

  [B] Photos
      Specify a folder and auto-extract timeline (EXIF metadata)

  [C] Social Media
      WhatsApp channels / Instagram exports

  [D] Upload Files
      PDF / screenshots / plain text

  [E] Paste Text
      Paste text directly

You can combine methods, or skip all and generate from manual info only.

Option A: iMessage / SMS / WhatsApp

iMessage file mode:

python3 ${CLAUDE_SKILL_DIR}/tools/imessage_parser.py --file {path} --target "{phone_or_name}" --output /tmp/imessage_out.txt

iMessage direct mode:

python3 ${CLAUDE_SKILL_DIR}/tools/imessage_parser.py --direct --target "{phone_or_name}" --output /tmp/imessage_out.txt

SMS:

python3 ${CLAUDE_SKILL_DIR}/tools/sms_parser.py --file {path} --target "{phone_or_name}" --output /tmp/sms_out.txt

WhatsApp:

python3 ${CLAUDE_SKILL_DIR}/tools/social_media_parser.py --file {path} --platform whatsapp --target "{name}" --output /tmp/whatsapp_out.txt

Option B: Photos

python3 ${CLAUDE_SKILL_DIR}/tools/photo_analyzer.py --dir {photo_directory} --output /tmp/photo_timeline.txt

Then Read /tmp/photo_timeline.txt.

Option C: Social Media

python3 ${CLAUDE_SKILL_DIR}/tools/social_media_parser.py \
  --file {path} \
  --platform {whatsapp|instagram|text} \
  --target "{name}" \
  --output /tmp/social_out.txt

Then Read /tmp/social_out.txt.

Option D: File Upload

• PDF / Images: Read
• Markdown / TXT: Read

Option E: Paste Text

Use user-pasted content directly.

If user says "no files" or "skip", generate from Step 1 info only.

Step 3: Analyze Source Material

Analyze in two tracks:

Track A (Memories)

• Use ${CLAUDE_SKILL_DIR}/prompts/memories_analyzer.md
• Extract timeline, routines, preferences, conflict patterns, emotional dynamics

Track B (Persona)

• Use ${CLAUDE_SKILL_DIR}/prompts/persona_analyzer.md
• Translate tags into behavior rules
• Extract communication style, emotional logic, relationship behavior

Step 4: Generate and Preview

Generate with:

• ${CLAUDE_SKILL_DIR}/prompts/memories_builder.md
• ${CLAUDE_SKILL_DIR}/prompts/persona_builder.md

Show both summaries and ask for confirmation.

Step 5: Write Files

After confirmation:

1. Create folders

mkdir -p crushes/{slug}/versions
mkdir -p crushes/{slug}/knowledge/chats
mkdir -p crushes/{slug}/knowledge/photos
mkdir -p crushes/{slug}/knowledge/social

2. Write memories.md

3. Write persona.md

4. Write meta.json

5. Generate full SKILL.md

Then tell the user:

Crush Skill created.

Location: crushes/{slug}/
Commands: /{slug} (full)
          /{slug}-memories
          /{slug}-persona

If anything feels wrong, say "she wouldn't do that" and I will update it.

---

Evolution Mode: Append Materials

When user provides new content:

1. Read new content via Step 2
2. Read existing crushes/{slug}/memories.md and persona.md
3. Use ${CLAUDE_SKILL_DIR}/prompts/merger.md
4. Backup current version:

   python3 ${CLAUDE_SKILL_DIR}/tools/version_manager.py --action backup --slug {slug} --base-dir ./crushes
   

5. Append updates with Edit
6. Regenerate SKILL.md
7. Update meta.json version and timestamp

---

Evolution Mode: Conversation Correction

When user says output is wrong:

1. Use ${CLAUDE_SKILL_DIR}/prompts/correction_handler.md
2. Route correction to memories or persona layer
3. Generate correction entry
4. Append to ## Correction Log
5. Regenerate SKILL.md

---

Management Commands

/list-crushes

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

/crush-rollback {slug} {version}

python3 ${CLAUDE_SKILL_DIR}/tools/version_manager.py --action rollback --slug {slug} --version {version} --base-dir ./crushes

/delete-crush {slug}

rm -rf crushes/{slug}