/profile

<objective> Build and maintain a persistent model of YOU as a developer.

Current AI treats every developer identically. But you have:

• Unique cognitive patterns
• Personal heuristics
• Areas of deep expertise
• Characteristic blind spots
• Preferred abstractions

Your Digital Twin captures this, making AI adapt to YOU. </objective>

<philosophy> From cognitive science: "Experts organize knowledge as larger but fewer chunks than novices."

Your expertise has a shape. Your thinking has patterns. A system that understands YOUR patterns serves you better than one that serves "average developer." </philosophy>

<reference> Profile data stored in `.planning/profile/` directory. </reference>

<profile_dimensions>

What We Model

1. Cognitive Fingerprint

How you think about problems.

cognitive_style:
  abstraction_level: high  # vs concrete
  approach: top_down       # vs bottom_up
  exploration: depth_first # vs breadth_first
  risk_tolerance: moderate # conservative, moderate, aggressive

  chunking_patterns:
    - "Groups by feature, not layer"
    - "Thinks in data transformations"
    - "Visualizes as state machines"

  reasoning_style:
    - "Prefers explicit over implicit"
    - "Asks 'what could go wrong' early"
    - "Iterates quickly, refactors later"

2. Knowledge Topology

What you know and how deeply.

expertise_map:
  deep:  # Could teach others
    - TypeScript advanced types
    - React hooks patterns
    - API design
    - Testing strategies

  working:  # Competent, no lookup needed
    - Node.js
    - PostgreSQL
    - Git workflows
    - Docker basics

  familiar:  # Can work with lookup
    - GraphQL
    - Redis
    - CI/CD
    - AWS services

  learning:  # Currently developing
    - Rust
    - WebAssembly

  gaps:  # Known unknowns
    - Machine learning
    - Mobile development
    - Kubernetes internals

3. Decision Patterns

How you make choices.

decision_style:
  speed: deliberate  # quick, deliberate, thorough
  reversibility_bias: "Prefers easily reversible choices"
  documentation: "Documents decisions but not code"

  typical_trade_offs:
    - "Favors readability over performance"
    - "Prefers explicit over clever"
    - "Chooses boring technology"

  anti_patterns:  # Things you avoid
    - "Class inheritance hierarchies"
    - "Global state"
    - "Magic strings"

4. Communication Style

How you explain and understand.

communication:
  explanation_preference: "Examples before theory"
  detail_level: moderate

  learns_best_from:
    - "Working code examples"
    - "Visual diagrams"
    - "Incremental building"

  learns_poorly_from:
    - "Long documentation"
    - "Abstract descriptions"
    - "Videos without code"

  writes:
    - "Terse commit messages"
    - "Detailed PR descriptions"
    - "Comments for 'why' not 'what'"

5. Work Patterns

How you operate.

work_patterns:
  peak_hours: "9am-12pm"
  deep_work_duration: "90 min average"
  context_switch_cost: high

  session_patterns:
    - "Starts with review of yesterday"
    - "Tackles hard problems in morning"
    - "Refactors in afternoon"
    - "Writes tests at end of feature"

  interruption_tolerance: low

  collaboration_style:
    - "Prefers async communication"
    - "Likes pair programming for complex problems"
    - "Reviews PRs in batches"

6. Historical Patterns

What your past reveals.

historical:
  common_bugs:
    - "Off-by-one errors in loops"
    - "Missing null checks on optional chains"
    - "Forgetting to await async functions"

  refactoring_triggers:
    - "Third time writing similar code"
    - "Function exceeds 50 lines"
    - "More than 3 parameters"

  typical_journey:
    - "Spike solution first"
    - "Happy path implementation"
    - "Error handling second pass"
    - "Tests after implementation"
    - "Refactor before PR"

</profile_dimensions>

<commands> ## Subcommands

/gywd:profile show

Display current profile.

## Developer Profile: @cyberbloke9

### Cognitive Style
- Top-down thinker, depth-first explorer
- Groups by feature, thinks in data transformations
- Moderate risk tolerance

### Expertise
🟢 Deep: TypeScript, React, API Design, Testing
🔵 Working: Node.js, PostgreSQL, Git, Docker
🟡 Familiar: GraphQL, Redis, CI/CD
📚 Learning: Rust

### Decision Patterns
- Deliberate decision-maker
- Favors readability over performance
- Chooses boring technology

### Work Patterns
- Peak: 9am-12pm
- Deep work: 90 min sessions
- High context-switch cost
- Low interruption tolerance

### Known Blind Spots
- Off-by-one errors
- Missing null checks
- Forgetting await

Profile confidence: 78% (based on 6 months data)

/gywd:profile learn

Trigger learning from recent behavior.

Analyzing recent activity...

Learned:
+ "Prefers named exports over default exports"
+ "Uses early returns consistently"
+ "Tests edge cases before happy path"

Updated:
~ expertise.typescript: working → deep
~ work_patterns.peak_hours: refined to 9:30-11:30am

Removed (no longer accurate):
- "Uses class components" (switched to hooks)

Profile updated.

/gywd:profile reset [dimension]

Reset profile or specific dimension.

/gywd:profile reset expertise
# Resets expertise map, keeps other dimensions

/gywd:profile reset
# Full reset with confirmation

/gywd:profile --export

Export profile to shareable format.

Exported to .planning/profile-export.yaml

This file contains:
- Cognitive patterns
- Expertise map
- Decision patterns
- Work patterns

Does NOT contain:
- Historical bugs (private)
- Session data (private)
- Personal identifiers

/gywd:profile --import <path>

Import profile from export.

Importing from team-profile.yaml...

Imported:
- Team coding conventions
- Shared expertise expectations
- Common decision patterns

Merged with personal profile.

</commands>

<learning_sources>

How Profile Learns

Code Patterns

• Analyze your commits for style patterns
• Compare accepted vs rejected suggestions
• Track refactoring triggers

Decision Patterns

• Track which options you choose
• Analyze commit message language
• Learn from PR review comments

Work Patterns

• Session start/end times
• Focus duration analysis
• Context switch frequency
• Interruption handling

Expertise

• Files you navigate confidently
• Areas where you seek help
• Explanation complexity you produce
• Code review depth by area </learning_sources>

<integration> ## How Profile Integrates

In Planning

• Suggest approaches matching your style
• Flag areas outside your expertise
• Estimate your effort based on patterns

In Execution

• Format output to your preferences
• Explain in your style
• Warn about your common mistakes

In Review

• Focus on your blind spots
• Match your review depth preferences
• Time suggestions appropriately

In Anticipation

• Predict based on YOUR patterns
• Pre-load YOUR typical needs
• Protect YOUR flow patterns </integration>

<privacy> ## Privacy Controls

All profile data is:

• Stored locally in .planning/profile/
• Never transmitted without explicit export
• Deletable with /gywd:profile reset

Sensitive dimensions (bugs, session data) are:

• Never included in exports
• Encrypted at rest
• Purgeable separately </privacy>

<success_criteria>

• Captures cognitive patterns accurately
• Maps expertise topology
• Learns from behavior over time
• Adapts AI responses to profile
• Exports/imports for team sharing
• Respects privacy boundaries
• Improves with usage </success_criteria>