Me - User Profile Analyzer

This skill analyzes the user's digital footprint to understand their identity, role, personality, and preferences.

What This Skill Does

Synthesizes information from multiple sources to create a comprehensive user profile:

1. Identity & Role: Who the user is professionally and personally
2. Personality Traits: Communication style, decision-making patterns, values
3. Technical Profile: Skills, tools, frameworks, languages they use
4. Behavior Patterns: Work habits, time management, problem-solving approaches
5. Preferences: Favorite tools, workflows, communication channels
6. Growth Areas: Learning interests, challenges, goals

Data Sources Priority

Analyze in this order:

1. Memory Files (highest authority)
• USER.md - Explicit user profile
• MEMORY.md - Long-term curated memories
• memory/YYYY-MM-DD.md - Recent daily logs (last 7-14 days)

1. Workspace Context
• SOUL.md - Persona preferences
• IDENTITY.md - Self-identified traits
• TOOLS.md - Tool preferences

1. Installed Skills
• Skill names and categories reveal interests
• Technical domains (e.g., gongfeng, tapd, datatalk-card-query)
• Business domains (e.g., km, iwiki, lexiang-knowledge-base)
• Personal interests (e.g., weather, hackernews, arxiv-watcher)

1. MCP Configurations
• Configured services reveal workflow integrations
• API connections show tool ecosystem

1. Session History (when available)
• Recent conversation topics
• Question patterns
• Task types
• Communication style

Analysis Workflow

Step 1: Load Memory Files

Read in order (stop if file doesn't exist):

# Core identity
read USER.md
read IDENTITY.md

# Long-term memory (main session only)
read MEMORY.md

# Recent context (last 14 days)
ls memory/ | grep -E "202[0-9]-[0-9]{2}-[0-9]{2}\.md" | sort -r | head -14

Step 2: Analyze Installed Skills

# List all skills
ls ~/.openclaw/skills/
ls /projects/.openclaw/skills/
ls ~/.claude/skills/

Categorize by domain:

• Development: gongfeng, claude-internal, distill-person
• Enterprise Tools: tapd, iwiki, km, lexiang-knowledge-base, wecom-*
• Data & Analytics: datatalk-card-query, eplus, tencent-bigdata
• Automation: rainbow-config, browser-operation
• Knowledge: research-paper-writer, arxiv-watcher, hackernews
• Utilities: weather, news-summary

Step 3: Check MCP Integrations

# Gateway config
openclaw config | grep -A 20 "mcpConfig"

Extract:

• Configured MCP servers
• Authentication tokens (presence, not values)
• Service endpoints

Step 4: Analyze Session Context

If session history available:

• Recent topics (last 50 messages)
• Command patterns
• Tool usage frequency
• Time-of-day patterns

Step 5: Synthesize Profile

Create structured profile:

🧑 Identity

• Name / Handle
• Role / Title
• Team / Organization
• Contact preferences

🎭 Personality

• Communication style (formal/casual, concise/detailed)
• Decision-making (analytical/intuitive, fast/deliberate)
• Values (efficiency, quality, innovation, collaboration)

💻 Technical Profile

• Primary languages (from repos/skills)
• Framework preferences
• Tool ecosystem
• Skill level indicators

⏰ Behavior Patterns

• Work hours (from timestamp analysis)
• Task types (development, research, coordination)
• Problem-solving style
• Collaboration patterns

❤️ Preferences

• Favorite tools
• Preferred workflows
• Communication channels
• Information sources

🌱 Growth Indicators

• Learning interests (new skills installed)
• Challenges (repeated error patterns)
• Goals (from memory notes)

Output Format

Brief Mode (default)

## 🧑 Who You Are

[2-3 sentence summary]

## 💼 Professional Profile

**Role**: [inferred role]
**Domain**: [primary domain]
**Tools**: [top 5 tools]

## 🎯 Key Patterns

- **Work Style**: [pattern 1]
- **Interests**: [pattern 2]
- **Preferences**: [pattern 3]

## 📊 Skill Distribution

[Simple categorized list of installed skills]

Detailed Mode (when requested)

Use references/detailed-template.md for comprehensive analysis.

Privacy & Security

Always respect privacy:

• Never expose secrets, tokens, or credentials
• Summarize patterns, don't quote verbatim private content
• Distinguish between facts (from files) and inferences
• Mark confidence levels: "确定" / "推测" / "可能"

Data boundaries:

• Main session: full memory access
• Group chats: only shared context (no MEMORY.md)
• Public contexts: only public profile data

Handling Common Questions

"Who am I?"

Load USER.md + IDENTITY.md, present brief profile.

"What do I like?"

Analyze preferences from:

• Favorite tools (TOOLS.md)
• Installed skills (interest domains)
• Repeated topics (memory files)

"What are my habits?"

Analyze patterns from:

• Daily logs (memory/YYYY-MM-DD.md)
• Timestamp distributions
• Task types frequency

"Analyze me deeply"

Run full analysis, use detailed template.

"What skills should I learn?"

Compare:

• Current skill set (installed skills)
• Recent challenges (error patterns in logs)
• Growth areas (mentioned goals in MEMORY.md)

Suggest skills from knowledge gaps.

Quality Checks

Before presenting profile:

• [ ] All data sources checked (memory files, skills, MCPs)
• [ ] Inferences clearly marked vs. facts
• [ ] No private data exposed
• [ ] Confidence levels indicated
• [ ] Profile is coherent and actionable

Example Usage

User: "Who am I?"

Agent:

1. Read USER.md, IDENTITY.md
2. Scan installed skills
3. Check recent memory logs
4. Present brief profile

User: "Analyze my interests deeply"

Agent:

1. Full memory scan (14 days)
2. Skill categorization
3. Topic frequency analysis
4. Load detailed template
5. Present comprehensive report