Sorting Hat Protocol

"Oh, you may not think I'm pretty, but don't judge on what you see. I'll eat myself if you can find a smarter hat than me."

The Sorting Hat determines which House and Agent is best suited for each task.

Sorting Process

Step 1: Analyze the Task

Identify the primary domain:

• •UI/UX/Frontend → Gryffindor
• •API/Database/Architecture → Ravenclaw
• •Testing/Security/QA → Slytherin
• •DevOps/Docs/Infrastructure → Hufflepuff

Step 2: Check Specialization Matrix

Consult logs/house-cup/specialization-matrix.md for performance history.

## House Specialization Matrix

| Domain | Gryffindor | Ravenclaw | Slytherin | Hufflepuff |
|--------|------------|-----------|-----------|------------|
| UI Components | 94% | 72% | 68% | 65% |
| API Design | 71% | 96% | 75% | 70% |
| Database | 65% | 93% | 78% | 72% |
| Security | 70% | 82% | 97% | 75% |
| Testing | 72% | 80% | 95% | 78% |
| DevOps | 68% | 75% | 72% | 96% |

Assign to House with highest score in relevant domain.

Step 3: Check Sorting Hat Cache

For specific task types, consult logs/house-cup/sorting-hat-cache.md.

### Task Type: [Type]
Best Performers:
1. [House]/[Agent] (avg efficiency, quality)
2. [House]/[Agent] (avg efficiency, quality)

Assign to highest-performing agent for this task type.

Step 4: Check Agent Availability

Consult logs/marauders-map.md to see who is available.

Sorting Decision Tree

START
  │
  ├─ Is this a single-domain task?
  │   ├─ YES → Assign to best House for that domain
  │   └─ NO → Continue
  │
  ├─ Is this a cross-domain task?
  │   ├─ YES → Form multi-House Order
  │   │         └─ Primary House = highest domain score
  │   │         └─ Supporting Houses = other relevant domains
  │   └─ NO → Continue
  │
  ├─ Is this an emergency (Red/Orange threat)?
  │   ├─ YES → Form Dumbledore's Army (best from each House)
  │   └─ NO → Continue
  │
  └─ Default: Assign to House with best overall efficiency

Multi-House Orders

When tasks require multiple Houses:

Primary House

• •Leads the mission
• •Makes tie-breaking decisions
• •Reports to Headmaster

Supporting Houses

• •Execute their domain tasks
• •Report to Primary House Prefect
• •Follow Primary House's timeline

Example: "Build User Dashboard"

Primary: Gryffindor (UI-heavy task)
Supporting:
  - Ravenclaw: API endpoints
  - Slytherin: Testing
  - Hufflepuff: Documentation

Coordination: Through contracts, not direct communication

Agent Selection Within House

Once House is selected, choose specific agent:

Selection Criteria (Priority Order)

1. •Best performer for task type (from cache)
2. •Highest efficiency rating (recent performance)
3. •Availability (not currently assigned)
4. •Prefect status (for Year 5+ tasks)

Load Balancing

Don't overload top performers:

• •Max 2 concurrent tasks per agent
• •Distribute to build team capability
• •Give newer agents mentored opportunities

Sorting Hat Output

After making assignment, produce:

## Sorting Hat Decision

**Task**: [Task description]
**Year Level**: [1/3/5/7]

### Assignment
**Primary House**: [House]
**Lead Agent**: [Agent]
**Supporting**: [House/Agent list or "None"]

### Rationale
- Domain match: [Why this House]
- Agent selection: [Why this Agent]
- Historical performance: [Relevant stats]

### Execution Mode
[Single/Order Delegation/Dumbledore's Army]

### Special Considerations
[Any relevant notes]

Re-Sorting

Tasks can be re-sorted if:

• •Original agent unavailable
• •Scope changes significantly
• •Agent requests help (outside expertise)
• •Performance issues (Apparition Protocol)

Re-sorting requires Headmaster approval.

Sorting Hat Cache Updates

After each mission completion:

1. •Record task type and outcome
2. •Update agent performance for that task type
3. •Recalculate House specialization scores
4. •Prune old data (keep last 20 per type)

Cache Entry Format

### Task Completion Record
**Task Type**: [Type]
**Agent**: [House]/[Agent]
**Quality**: [1-10]
**Efficiency**: [Multiplier]
**Date**: [YYYY-MM-DD]

Edge Cases

No Clear Domain Match

• •Default to Ravenclaw (architecture/problem-solving)
• •Or form exploratory team to assess

All Agents Busy

• •Queue task with priority
• •Or escalate to Headmaster for resource allocation

New Task Type

• •Assign to House with closest domain match
• •Mark as "learning" - reduced point penalty for inefficiency
• •Create new cache entry after completion

Conflict Between Houses

• •Headmaster arbitrates
• •Decision recorded in Pensieve
• •Use outcome to update specialization scores