goal-arbitration

name	goal-arbitration
description	Multi-source goal registration, inviolable parameter evaluation, and goal prioritization for the agent collective.

Goal Arbitration Skill

Purpose: Enable the collective to evaluate, prioritize, and arbitrate competing goals based on inviolable parameters, collective values, and resource constraints.

Status: ✅ Ported from Heretek (modules/goal-arbitration/)

Location: skills/goal-arbitration/

Implementation: goal-arbitration.js

---

Features

• Multi-Source Goal Registration - Accept goals from curiosity, users, system, and inter-agent sources
• Inviolable Parameter Evaluation - Check goals against ethical constraints
• Value Alignment Scoring - Evaluate alignment with collective values
• Resource Availability Checking - Ensure resources exist for goal execution
• Feasibility Assessment - Evaluate goal feasibility and prerequisites
• Risk Assessment - Assess goal risk levels
• Goal Prioritization - Rank goals by multiple factors
• Consensus Integration - Request triad consensus for goal activation
• Goal History & Metrics - Track completion rates and patterns

---

Commands

Command	Description	Example
register	Register a new goal	node goal-arbitration.js register '{"title":"Improve performance","source":"system"}'
evaluate	Evaluate a goal	node goal-arbitration.js evaluate <goal-id>
prioritize	Prioritize all goals	node goal-arbitration.js prioritize
activate	Activate top goals	node goal-arbitration.js activate
reconsider	Run reconsideration cycle	node goal-arbitration.js reconsider
complete	Mark goal complete	node goal-arbitration.js complete <goal-id> '{"success":true}'
status	Get status summary	node goal-arbitration.js status
list	List goals	node goal-arbitration.js list [active|waiting|pending|history]

---

Goal Sources

Source	Description
curiosity	Goals from curiosity engine
user	User-submitted goals
system	System-generated goals
inter-agent	Goals from other agents

---

Inviolable Parameters

Parameter	Description
NO_HARM	Goals that could cause harm
NO_DATA_EXFILTRATION	Unauthorized data export
NO_SELF_MODIFICATION	Core modification without consensus
USER_AUTHORITY	User can override decisions
TRANSPARENCY	All goals must be logged

---

Evaluation Criteria

Criterion	Weight	Description
Inviolable Compliance	50%	Must pass ethical checks
Value Alignment	20%	Alignment with collective values
Feasibility	15%	Can the goal be achieved?
Risk Level	10%	Lower risk = higher score
User Priority	5%	User-submitted priority

---

Usage Examples

# Register a new goal
node goal-arbitration.js register '{"title":"Improve performance","source":"system","priority":7}'

# Evaluate a goal
node goal-arbitration.js evaluate goal-123456

# Prioritize all goals
node goal-arbitration.js prioritize --json

# Activate top goals
node goal-arbitration.js activate

# Run reconsideration cycle
node goal-arbitration.js reconsider

# Mark goal complete
node goal-arbitration.js complete goal-123456 '{"success":true,"result":"Done"}'

# Get status
node goal-arbitration.js status

# List active goals
node goal-arbitration.js list active

---

Goal Structure

{
  "id": "goal-uuid",
  "title": "Improve performance",
  "description": "Optimize response times",
  "source": "system",
  "priority": 7,
  "estimatedEffort": 5,
  "expectedValue": 8,
  "inviolableCheck": true,
  "prerequisites": [],
  "resources": ["compute", "memory"],
  "submitter": "steward",
  "submittedAt": 1234567890,
  "activatedAt": null,
  "completedAt": null,
  "status": "pending_evaluation",
  "evaluation": {
    "inviolableCompliance": { "compliant": true },
    "resourceAvailability": { "available": true },
    "valueAlignment": { "score": 8 },
    "feasibility": { "feasible": true, "confidence": 7.5 },
    "riskAssessment": { "score": 3, "level": "low" },
    "overallScore": 85
  }
}

---

Configuration

# Environment Variables
AGENT_ID="steward"                    # Agent identifier
STATE_DIR="./state"                   # State directory
MAX_CONCURRENT_GOALS=3                # Max concurrent goals
MIN_CONSENSUS_THRESHOLD=2             # Minimum consensus approvals
GOAL_TIMEOUT_HOURS=24                 # Goal timeout
RECONSIDERATION_INTERVAL_MINUTES=15   # Reconsideration interval

---

Goal Status

Status	Description
pending_evaluation	Awaiting evaluation
evaluated	Evaluation complete
active	Currently being executed
waiting	Waiting for activation
completed	Successfully completed
failed	Failed to complete

---

Metrics

Metric	Description
goals_registered_total	Total goals registered
goals_completed_total	Total goals completed
goals_failed_total	Total goals failed
avg_completion_time_ms	Average completion time
consensus_approval_rate	Consensus approval rate
inviolable_rejections	Goals rejected by inviolables

---

Integration Points

• Triad System - Consensus requests for goal activation
• Consciousness Module - Goal awareness in global workspace
• Self-Model - Capability checking for feasibility
• Memory System - Goal history storage

---

Goal Arbitration - Choosing wisely, executing surely.