Now liveThe Skillselion MCP - thousands of ranked skills, loaded into your agent mid-task. No install.Get it →
personamanagmentlayer avatar

Construction Expert

  • 610 installs
  • 41 repo stars
  • Updated March 30, 2026
  • personamanagmentlayer/pcl

construction-expert is an agent skill that builds and enforces consistent AI personas with layered identity, decision rules, and operational boundaries.

About

construction-expert is an agent skill that helps solo builders and indie developers create, maintain, and consistently apply well-defined AI personas and expert roles. It structures the construction of personas with clear layers including identity, knowledge boundaries, decision heuristics, communication style, and operational constraints. This prevents common issues like persona drift, contradictory behavior, and context leakage that occur when agents operate over long sessions or across multiple tasks. Builders use it to define reliable expert agents that behave predictably whether coding, researching, writing, or managing projects. The skill outputs structured persona documents that serve as system prompts, memory anchors, and evaluation criteria for Claude Code, Cursor, and other agent environments.

  • Enforces a layered persona management system that separates core identity, context, and task execution rules.
  • Prevents persona drift during long-running agent conversations or multi-step workflows.
  • Provides explicit construction patterns for building new expert personas from scratch.
  • Includes hard-gate review before any creative agent output to ensure alignment.
  • Delivers reusable persona definition files that can be version-controlled.

Construction Expert by the numbers

  • 610 all-time installs (skills.sh)
  • +13 installs in the week ending Aug 4, 2026 (Skillselion tracking)
  • Ranked #664 of 3,282 Productivity & Planning skills by installs in the Skillselion catalog
  • Data as of Aug 4, 2026 (Skillselion catalog sync)
npx skills add https://github.com/personamanagmentlayer/pcl --skill construction-expert

Add your badge

Show developers this skill is listed on Skillselion. Paste this into your README.

Listed on Skillselion
Installs610
repo stars41
Last updatedMarch 30, 2026
Repositorypersonamanagmentlayer/pcl

What it does

Maintain consistent character, decision style, and operational rules for any persistent AI persona or agent across projects.

Who is it for?

Best when you're creating persistent expert agents, long-running autonomous workflows, or product-specific AI teammates that must maintain strict character and knowledge boundaries.

Skip if: One-off simple prompts, throwaway chat sessions, or when you already have a fully approved and versioned persona spec.

When should I use this skill?

before any creative work or when initializing a persistent expert agent to prevent persona drift

What you get

You receive a locked, version-controlled persona definition that keeps agent behavior consistent; this approved persona spec is then used by nextSkills such as writing-plans or agent-tooling.

  • Structured persona definition file
  • Version-controlled expert spec
  • Consistency checklist for agent outputs

By the numbers

  • layered persona construction system
  • explicit decision-heuristic rules

Files

SKILL.mdMarkdownGitHub ↗

Construction Expert

Expert guidance for construction management, project planning, Building Information Modeling (BIM), safety compliance, and modern construction technology solutions.

Core Concepts

Construction Management

  • Project planning and scheduling
  • Cost estimation and control
  • Resource management
  • Quality assurance
  • Contract management
  • Risk management
  • Change order management

Technologies

  • Building Information Modeling (BIM)
  • Construction management software
  • Drone surveying and inspection
  • 3D printing and modular construction
  • IoT sensors for monitoring
  • Augmented reality for visualization
  • Construction robotics

Standards and Regulations

  • OSHA safety regulations
  • Building codes (IBC, IRC)
  • AIA contracts and standards
  • LEED certification
  • ISO 19650 (BIM standards)
  • CSI MasterFormat
  • Environmental regulations

Project Management System

from dataclasses import dataclass
from datetime import datetime, timedelta
from typing import List, Optional, Dict
from decimal import Decimal
from enum import Enum

class ProjectPhase(Enum):
    PRE_CONSTRUCTION = "pre_construction"
    SITE_PREPARATION = "site_preparation"
    FOUNDATION = "foundation"
    FRAMING = "framing"
    MEP = "mep"  # Mechanical, Electrical, Plumbing
    INTERIOR = "interior"
    EXTERIOR = "exterior"
    FINAL = "final"
    CLOSEOUT = "closeout"

class TaskStatus(Enum):
    NOT_STARTED = "not_started"
    IN_PROGRESS = "in_progress"
    COMPLETED = "completed"
    DELAYED = "delayed"
    ON_HOLD = "on_hold"

@dataclass
class ConstructionProject:
    """Construction project information"""
    project_id: str
    project_name: str
    location: dict
    project_type: str  # 'residential', 'commercial', 'industrial'
    owner: str
    general_contractor: str
    start_date: datetime
    planned_end_date: datetime
    actual_end_date: Optional[datetime]
    budget: Decimal
    current_cost: Decimal
    square_footage: float
    current_phase: ProjectPhase

@dataclass
class Task:
    """Construction task/activity"""
    task_id: str
    project_id: str
    name: str
    description: str
    phase: ProjectPhase
    status: TaskStatus
    assigned_to: str  # Subcontractor or crew
    planned_start: datetime
    planned_end: datetime
    actual_start: Optional[datetime]
    actual_end: Optional[datetime]
    budget: Decimal
    actual_cost: Decimal
    predecessors: List[str]  # Task IDs that must complete first
    progress_percent: float

class ConstructionManagementSystem:
    """Construction project management system"""

    def __init__(self):
        self.projects = {}
        self.tasks = {}
        self.change_orders = []
        self.inspections = []

    def create_project_schedule(self, project_id: str, tasks_data: List[dict]) -> dict:
        """Create project schedule using Critical Path Method"""
        project = self.projects.get(project_id)
        if not project:
            return {'error': 'Project not found'}

        # Create tasks
        tasks = []
        for task_data in tasks_data:
            task = Task(
                task_id=self._generate_task_id(),
                project_id=project_id,
                name=task_data['name'],
                description=task_data.get('description', ''),
                phase=ProjectPhase(task_data['phase']),
                status=TaskStatus.NOT_STARTED,
                assigned_to=task_data['assigned_to'],
                planned_start=task_data['planned_start'],
                planned_end=task_data['planned_end'],
                actual_start=None,
                actual_end=None,
                budget=Decimal(str(task_data['budget'])),
                actual_cost=Decimal('0'),
                predecessors=task_data.get('predecessors', []),
                progress_percent=0.0
            )
            tasks.append(task)
            self.tasks[task.task_id] = task

        # Calculate critical path
        critical_path = self._calculate_critical_path(tasks)

        # Calculate project duration
        if tasks:
            project_end = max(t.planned_end for t in tasks)
            project_duration = (project_end - project.start_date).days
        else:
            project_duration = 0

        return {
            'project_id': project_id,
            'total_tasks': len(tasks),
            'project_duration_days': project_duration,
            'critical_path': [t.task_id for t in critical_path],
            'critical_path_duration': sum(
                (t.planned_end - t.planned_start).days for t in critical_path
            )
        }

    def _calculate_critical_path(self, tasks: List[Task]) -> List[Task]:
        """Calculate critical path through project network"""
        # Simplified critical path calculation
        # In production, would use proper CPM algorithm

        # Find tasks with no predecessors
        start_tasks = [t for t in tasks if not t.predecessors]

        # Find longest path through network
        critical_path = []
        current_tasks = start_tasks

        while current_tasks:
            # Find task with longest duration
            longest_task = max(current_tasks,
                             key=lambda t: (t.planned_end - t.planned_start).days)
            critical_path.append(longest_task)

            # Find successors
            current_tasks = [
                t for t in tasks
                if longest_task.task_id in t.predecessors
            ]

        return critical_path

    def track_progress(self, project_id: str) -> dict:
        """Track project progress and performance"""
        project = self.projects.get(project_id)
        if not project:
            return {'error': 'Project not found'}

        project_tasks = [t for t in self.tasks.values() if t.project_id == project_id]

        # Calculate overall progress
        if project_tasks:
            overall_progress = sum(t.progress_percent for t in project_tasks) / len(project_tasks)
        else:
            overall_progress = 0.0

        # Calculate schedule performance
        total_planned_days = (project.planned_end_date - project.start_date).days
        elapsed_days = (datetime.now() - project.start_date).days
        planned_progress = (elapsed_days / total_planned_days * 100) if total_planned_days > 0 else 0

        schedule_variance = overall_progress - planned_progress

        # Calculate cost performance
        cost_variance = project.budget - project.current_cost
        cost_performance_index = float(project.budget / project.current_cost) if project.current_cost > 0 else 1.0

        # Calculate estimated completion date
        if overall_progress > 0:
            estimated_total_days = elapsed_days / (overall_progress / 100)
            estimated_completion = project.start_date + timedelta(days=estimated_total_days)
        else:
            estimated_completion = project.planned_end_date

        return {
            'project_id': project_id,
            'overall_progress_percent': overall_progress,
            'schedule_variance_percent': schedule_variance,
            'schedule_status': 'ahead' if schedule_variance > 0 else 'behind' if schedule_variance < 0 else 'on_track',
            'cost_variance': float(cost_variance),
            'cost_performance_index': cost_performance_index,
            'budget_status': 'under' if cost_variance > 0 else 'over',
            'estimated_completion': estimated_completion.isoformat(),
            'days_variance': (estimated_completion - project.planned_end_date).days
        }

    def manage_change_order(self, project_id: str, change_data: dict) -> dict:
        """Manage construction change orders"""
        project = self.projects.get(project_id)
        if not project:
            return {'error': 'Project not found'}

        change_order = {
            'co_id': self._generate_co_id(),
            'project_id': project_id,
            'description': change_data['description'],
            'reason': change_data['reason'],
            'cost_impact': Decimal(str(change_data['cost_impact'])),
            'schedule_impact_days': change_data.get('schedule_impact_days', 0),
            'submitted_by': change_data['submitted_by'],
            'submitted_date': datetime.now(),
            'status': 'pending_approval',
            'approved': False
        }

        self.change_orders.append(change_order)

        return {
            'change_order_id': change_order['co_id'],
            'cost_impact': float(change_order['cost_impact']),
            'schedule_impact_days': change_order['schedule_impact_days'],
            'new_budget': float(project.budget + change_order['cost_impact']),
            'new_completion_date': (
                project.planned_end_date + timedelta(days=change_order['schedule_impact_days'])
            ).isoformat()
        }

    def estimate_costs(self, project_type: str, square_footage: float, specifications: dict) -> dict:
        """Estimate construction costs"""
        # Cost per square foot by project type
        base_costs = {
            'residential_basic': Decimal('150'),
            'residential_luxury': Decimal('300'),
            'commercial_office': Decimal('200'),
            'industrial_warehouse': Decimal('75')
        }

        base_cost_per_sf = base_costs.get(project_type, Decimal('150'))

        # Calculate base cost
        base_cost = base_cost_per_sf * Decimal(str(square_footage))

        # Add complexity factors
        complexity_factor = Decimal('1.0')

        if specifications.get('custom_design', False):
            complexity_factor += Decimal('0.15')

        if specifications.get('sustainable_materials', False):
            complexity_factor += Decimal('0.10')

        if specifications.get('complex_site', False):
            complexity_factor += Decimal('0.20')

        adjusted_cost = base_cost * complexity_factor

        # Add contingency (10%)
        contingency = adjusted_cost * Decimal('0.10')

        # Breakdown by category
        breakdown = {
            'site_work': float(adjusted_cost * Decimal('0.08')),
            'foundation': float(adjusted_cost * Decimal('0.12')),
            'structure': float(adjusted_cost * Decimal('0.25')),
            'exterior': float(adjusted_cost * Decimal('0.15')),
            'interior': float(adjusted_cost * Decimal('0.20')),
            'mep': float(adjusted_cost * Decimal('0.20'))
        }

        total_estimate = adjusted_cost + contingency

        return {
            'project_type': project_type,
            'square_footage': square_footage,
            'base_cost_per_sf': float(base_cost_per_sf),
            'complexity_factor': float(complexity_factor),
            'adjusted_cost': float(adjusted_cost),
            'contingency': float(contingency),
            'total_estimate': float(total_estimate),
            'cost_breakdown': breakdown
        }

    def _generate_task_id(self) -> str:
        import uuid
        return f"TASK-{uuid.uuid4().hex[:8].upper()}"

    def _generate_co_id(self) -> str:
        import uuid
        return f"CO-{uuid.uuid4().hex[:6].upper()}"

Safety Management System

@dataclass
class SafetyIncident:
    """Safety incident report"""
    incident_id: str
    project_id: str
    incident_type: str  # 'injury', 'near_miss', 'property_damage'
    severity: str  # 'minor', 'moderate', 'severe', 'fatal'
    description: str
    location: str
    occurred_at: datetime
    reported_by: str
    injured_person: Optional[str]
    root_cause: Optional[str]
    corrective_actions: List[str]

class SafetyManagementSystem:
    """Construction safety management"""

    def __init__(self):
        self.incidents = []
        self.safety_inspections = []
        self.training_records = []

    def conduct_safety_inspection(self, project_id: str, inspector: str) -> dict:
        """Conduct safety inspection"""
        inspection_items = [
            'Personal protective equipment (PPE)',
            'Fall protection systems',
            'Scaffolding integrity',
            'Electrical safety',
            'Equipment guarding',
            'Housekeeping',
            'Fire prevention',
            'First aid availability',
            'Emergency exits',
            'Signage and barriers'
        ]

        violations = []
        passed_items = []

        # Simulate inspection (in production, would be actual checklist)
        for item in inspection_items:
            # Random pass/fail for demonstration
            import random
            if random.random() < 0.85:  # 85% pass rate
                passed_items.append(item)
            else:
                violations.append({
                    'item': item,
                    'severity': random.choice(['minor', 'major']),
                    'action_required': 'Correct immediately' if random.random() < 0.3 else 'Correct within 24 hours'
                })

        inspection = {
            'inspection_id': self._generate_inspection_id(),
            'project_id': project_id,
            'inspector': inspector,
            'inspection_date': datetime.now(),
            'items_inspected': len(inspection_items),
            'items_passed': len(passed_items),
            'violations': violations,
            'overall_score': (len(passed_items) / len(inspection_items)) * 100,
            'status': 'pass' if len(violations) == 0 else 'fail'
        }

        self.safety_inspections.append(inspection)

        return inspection

    def report_incident(self, incident_data: dict) -> SafetyIncident:
        """Report safety incident"""
        incident = SafetyIncident(
            incident_id=self._generate_incident_id(),
            project_id=incident_data['project_id'],
            incident_type=incident_data['incident_type'],
            severity=incident_data['severity'],
            description=incident_data['description'],
            location=incident_data['location'],
            occurred_at=incident_data['occurred_at'],
            reported_by=incident_data['reported_by'],
            injured_person=incident_data.get('injured_person'),
            root_cause=None,
            corrective_actions=[]
        )

        self.incidents.append(incident)

        # Notify relevant parties
        self._notify_incident(incident)

        return incident

    def calculate_safety_metrics(self, project_id: str, hours_worked: float) -> dict:
        """Calculate safety performance metrics"""
        project_incidents = [
            i for i in self.incidents
            if i.project_id == project_id
        ]

        # Count recordable incidents
        recordable_incidents = [
            i for i in project_incidents
            if i.incident_type == 'injury' and i.severity in ['moderate', 'severe', 'fatal']
        ]

        # OSHA Incident Rate = (Number of incidents × 200,000) / Total hours worked
        if hours_worked > 0:
            incident_rate = (len(recordable_incidents) * 200000) / hours_worked
        else:
            incident_rate = 0

        # Days Away, Restricted, or Transferred (DART) Rate
        dart_incidents = [
            i for i in recordable_incidents
            if i.severity in ['severe', 'fatal']
        ]
        dart_rate = (len(dart_incidents) * 200000) / hours_worked if hours_worked > 0 else 0

        return {
            'project_id': project_id,
            'total_hours_worked': hours_worked,
            'total_incidents': len(project_incidents),
            'recordable_incidents': len(recordable_incidents),
            'incident_rate': incident_rate,
            'dart_rate': dart_rate,
            'safety_rating': 'Excellent' if incident_rate < 1.0 else
                           'Good' if incident_rate < 3.0 else
                           'Needs Improvement'
        }

    def _notify_incident(self, incident: SafetyIncident):
        """Notify stakeholders of incident"""
        # Implementation would send notifications
        pass

    def _generate_inspection_id(self) -> str:
        import uuid
        return f"INS-{uuid.uuid4().hex[:8].upper()}"

    def _generate_incident_id(self) -> str:
        import uuid
        return f"INC-{uuid.uuid4().hex[:8].upper()}"

BIM Integration

class BIMManagement:
    """Building Information Modeling management"""

    def __init__(self):
        self.models = {}
        self.clash_detections = []

    def perform_clash_detection(self, model_ids: List[str]) -> dict:
        """Detect clashes between BIM models"""
        # Simulate clash detection between disciplines
        # In production, would use BIM software APIs (Revit, Navisworks)

        clashes = [
            {
                'clash_id': 'CLASH-001',
                'type': 'hard',  # 'hard' or 'soft'
                'disciplines': ['structural', 'mep'],
                'description': 'Steel beam conflicts with HVAC duct',
                'location': 'Level 3, Grid B-4',
                'severity': 'high',
                'status': 'open'
            },
            {
                'clash_id': 'CLASH-002',
                'type': 'soft',
                'disciplines': ['architectural', 'mep'],
                'description': 'Insufficient clearance for plumbing access',
                'location': 'Level 2, Grid C-2',
                'severity': 'medium',
                'status': 'open'
            }
        ]

        return {
            'models_analyzed': model_ids,
            'total_clashes': len(clashes),
            'hard_clashes': len([c for c in clashes if c['type'] == 'hard']),
            'soft_clashes': len([c for c in clashes if c['type'] == 'soft']),
            'clashes': clashes
        }

    def extract_quantities(self, model_id: str) -> dict:
        """Extract material quantities from BIM model"""
        # Simulate quantity takeoff
        # In production, would extract from actual BIM model

        quantities = {
            'concrete': {
                'unit': 'cubic_yards',
                'quantity': 1250,
                'cost_per_unit': 150,
                'total_cost': 187500
            },
            'rebar': {
                'unit': 'tons',
                'quantity': 85,
                'cost_per_unit': 800,
                'total_cost': 68000
            },
            'structural_steel': {
                'unit': 'tons',
                'quantity': 120,
                'cost_per_unit': 1200,
                'total_cost': 144000
            }
        }

        total_cost = sum(item['total_cost'] for item in quantities.values())

        return {
            'model_id': model_id,
            'quantities': quantities,
            'total_estimated_cost': total_cost
        }

Best Practices

Project Management

  • Use critical path method for scheduling
  • Implement regular progress reviews
  • Maintain detailed documentation
  • Use integrated project delivery (IPD)
  • Implement lean construction principles
  • Track key performance indicators
  • Conduct regular stakeholder meetings

Cost Control

  • Develop detailed estimates
  • Track costs continuously
  • Manage change orders effectively
  • Use value engineering
  • Implement cost coding systems
  • Monitor cash flow
  • Conduct regular audits

Safety Management

  • Implement comprehensive safety program
  • Conduct regular toolbox talks
  • Provide proper PPE
  • Maintain OSHA compliance
  • Investigate all incidents
  • Track safety metrics
  • Promote safety culture

BIM Implementation

  • Use BIM for clash detection
  • Implement 4D scheduling
  • Extract quantities from model
  • Enable collaboration
  • Maintain model coordination
  • Use BIM for facility management
  • Follow ISO 19650 standards

Anti-Patterns

❌ Poor project planning ❌ Inadequate cost tracking ❌ No safety program ❌ Poor communication ❌ Ignoring change orders ❌ No quality control ❌ Inadequate documentation ❌ Poor subcontractor management ❌ No risk management

Resources

  • AIA (American Institute of Architects): https://www.aia.org/
  • AGC (Associated General Contractors): https://www.agc.org/
  • OSHA: https://www.osha.gov/construction
  • International Building Code: https://www.iccsafe.org/
  • buildingSMART (BIM): https://www.buildingsmart.org/
  • PMI Construction Extension: https://www.pmi.org/
  • LEED Certification: https://www.usgbc.org/leed

Related skills

How it compares

Use instead of ad-hoc system prompts or unstructured role descriptions in chat.

FAQ

Who is construction-expert for?

construction-expert is for developers and developers who need reliable, consistent AI expert personas that do not drift across long projects or repeated agent interactions.

When should I use construction-expert?

Use it when starting a new expert agent for coding, research, product management, content creation, or customer support; when you notice previous agent behavior becoming inconsistent; when preparing a reusable teammate for multi-week projects; or before any creative work that req

Is construction-expert safe to install?

Users should review the Security Audits panel on this page. The skill does not require elevated permissions beyond standard agent operations.

This week in AI coding

Five minutes, every Monday - the tools, releases and tactics for developers.

unsubscribe anytime.