Verification Loop Construction Agent Skill

Comprehensive verification system for construction automation deliverables. Use after completing estimates, schedules, reports, or data processing tasks to ensure quality.

5k tokens

context cost

the whole folder, loaded on every use

files

instructions only

copies elsewhere

how many repositories repackaged it

264

stars on the repo

on the repository, not the skill itself

Install

one command, takes just this skill from the repository

npx skills add https://github.com/datadrivenconstruction/DDC_Skills_for_AI_Agents_in_Construction --skill verification-loop-construction

The instruction itself

17 sections, as written by the author

Verification Loop for Construction Automation

A systematic verification framework ensuring quality of construction automation outputs before delivery or deployment.

When to Use

Invoke this skill:

After generating cost estimates
After creating or updating schedules
After processing BIM/CAD data
After generating reports (daily, weekly, monthly)
After running data pipelines
Before submitting documents to clients
Before deploying automation workflows

Verification Phases

Phase 1: Data Integrity Check

def verify_data_integrity(output: dict) -> VerificationResult:
    """Check data completeness and consistency"""

    checks = []

    # Completeness check
    required_fields = get_required_fields(output['type'])
    missing = [f for f in required_fields if f not in output]
    checks.append({
        'name': 'Completeness',
        'status': 'PASS' if not missing else 'FAIL',
        'details': f'Missing fields: {missing}' if missing else 'All required fields present'
    })

    # Consistency check
    inconsistencies = find_inconsistencies(output)
    checks.append({
        'name': 'Consistency',
        'status': 'PASS' if not inconsistencies else 'WARN',
        'details': inconsistencies or 'No inconsistencies found'
    })

    # Referential integrity
    broken_refs = check_references(output)
    checks.append({
        'name': 'Referential Integrity',
        'status': 'PASS' if not broken_refs else 'FAIL',
        'details': f'Broken references: {broken_refs}' if broken_refs else 'All references valid'
    })

    return VerificationResult(checks)

Data Integrity Checklist

[ ] All required fields populated
[ ] No duplicate records
[ ] Foreign keys resolve correctly
[ ] Date formats consistent
[ ] Currency values formatted correctly
[ ] Units of measure standardized

Phase 2: Business Logic Verification

def verify_business_logic(output: dict) -> VerificationResult:
    """Verify construction-specific business rules"""

    checks = []

    # Cost estimate checks
    if output['type'] == 'cost_estimate':
        # Verify totals match line items
        calculated_total = sum(item['amount'] for item in output['line_items'])
        declared_total = output['total']
        variance = abs(calculated_total - declared_total)

        checks.append({
            'name': 'Total Accuracy',
            'status': 'PASS' if variance < 0.01 else 'FAIL',
            'details': f'Calculated: {calculated_total}, Declared: {declared_total}'
        })

        # Verify markup applied correctly
        for item in output['line_items']:
            expected_markup = item['base_cost'] * (1 + item['markup_rate'])
            if abs(item['amount'] - expected_markup) > 0.01:
                checks.append({
                    'name': f'Markup Check - {item["id"]}',
                    'status': 'FAIL',
                    'details': f'Expected: {expected_markup}, Got: {item["amount"]}'
                })

    # Schedule checks
    if output['type'] == 'schedule':
        # Verify dependencies
        for task in output['tasks']:
            for pred_id in task.get('predecessors', []):
                pred = find_task(output['tasks'], pred_id)
                if pred and pred['end_date'] > task['start_date']:
                    checks.append({
                        'name': f'Dependency Violation - {task["id"]}',
                        'status': 'FAIL',
                        'details': f'Task starts before predecessor {pred_id} ends'
                    })

        # Verify resource allocation
        resource_conflicts = find_resource_conflicts(output['tasks'])
        checks.append({
            'name': 'Resource Conflicts',
            'status': 'PASS' if not resource_conflicts else 'WARN',
            'details': resource_conflicts or 'No resource conflicts'
        })

    return VerificationResult(checks)

Business Logic Checklist

For Cost Estimates:

[ ] Line item totals = declared subtotals
[ ] Subtotals + markups = grand total
[ ] Tax calculations correct
[ ] Contingency applied consistently
[ ] Unit prices within expected ranges
[ ] Quantities match QTO

For Schedules:

[ ] Dependencies respected (no backwards links)
[ ] Critical path calculated correctly
[ ] No resource over-allocation
[ ] Milestones properly dated
[ ] Working days/calendar correct
[ ] Float values logical

For BIM Data:

[ ] Element counts match expected
[ ] Property values within ranges
[ ] Spatial relationships valid
[ ] Level associations correct
[ ] Classification codes valid

Phase 3: Format & Standards Compliance

def verify_standards_compliance(output: dict) -> VerificationResult:
    """Verify compliance with construction standards"""

    checks = []

    # CSI classification check
    if 'csi_codes' in output:
        invalid_codes = []
        for code in output['csi_codes']:
            if not validate_csi_code(code):
                invalid_codes.append(code)

        checks.append({
            'name': 'CSI Code Validation',
            'status': 'PASS' if not invalid_codes else 'WARN',
            'details': f'Invalid codes: {invalid_codes}' if invalid_codes else 'All codes valid'
        })

    # CWICR mapping check
    if output.get('cwicr_mapped'):
        unmapped = [item for item in output['items'] if not item.get('cwicr_id')]
        checks.append({
            'name': 'CWICR Mapping',
            'status': 'PASS' if not unmapped else 'WARN',
            'details': f'{len(unmapped)} items unmapped' if unmapped else 'All items mapped'
        })

    # Document format check
    if output['type'] == 'report':
        format_issues = validate_report_format(output)
        checks.append({
            'name': 'Report Format',
            'status': 'PASS' if not format_issues else 'WARN',
            'details': format_issues or 'Format compliant'
        })

    return VerificationResult(checks)

Standards Compliance Checklist

[ ] CSI MasterFormat codes valid (6-digit)
[ ] UniFormat codes valid (if used)
[ ] CWICR mappings present
[ ] ISO dates (YYYY-MM-DD)
[ ] Currency format consistent
[ ] Measurement units consistent (metric/imperial)

Phase 4: Output Quality Check

def verify_output_quality(output: dict) -> VerificationResult:
    """Check output quality and presentation"""

    checks = []

    # Formatting check
    if output['format'] == 'excel':
        checks.extend([
            {
                'name': 'Column Headers',
                'status': check_headers_present(output),
                'details': 'Headers in first row'
            },
            {
                'name': 'Number Formatting',
                'status': check_number_format(output),
                'details': 'Currencies and percentages formatted'
            },
            {
                'name': 'Print Area',
                'status': check_print_area(output),
                'details': 'Print area set for clean output'
            }
        ])

    if output['format'] == 'pdf':
        checks.extend([
            {
                'name': 'Page Layout',
                'status': check_page_layout(output),
                'details': 'Margins and orientation correct'
            },
            {
                'name': 'Images Rendered',
                'status': check_images(output),
                'details': 'All images/charts visible'
            },
            {
                'name': 'Fonts Embedded',
                'status': check_fonts(output),
                'details': 'Fonts embedded for portability'
            }
        ])

    # Data visualization check
    if 'charts' in output:
        for chart in output['charts']:
            checks.append({
                'name': f'Chart - {chart["title"]}',
                'status': validate_chart(chart),
                'details': 'Labels, legends, and data visible'
            })

    return VerificationResult(checks)

Output Quality Checklist

[ ] Headers/footers present
[ ] Page numbers correct
[ ] Company branding applied
[ ] Charts readable
[ ] Tables properly aligned
[ ] Links/references working

Phase 5: Cross-Reference Validation

def verify_cross_references(output: dict, sources: list) -> VerificationResult:
    """Validate output against source data"""

    checks = []

    for source in sources:
        # Compare key metrics
        metrics = extract_comparable_metrics(output, source)

        for metric_name, (output_val, source_val) in metrics.items():
            variance_pct = abs(output_val - source_val) / source_val * 100 if source_val else 0

            status = 'PASS'
            if variance_pct > 5:
                status = 'WARN'
            if variance_pct > 10:
                status = 'FAIL'

            checks.append({
                'name': f'{metric_name} vs {source["name"]}',
                'status': status,
                'details': f'Output: {output_val}, Source: {source_val}, Variance: {variance_pct:.1f}%'
            })

    return VerificationResult(checks)

Cross-Reference Checklist

[ ] QTO matches BIM model
[ ] Estimate matches specifications
[ ] Schedule aligns with estimate
[ ] Budget matches contract
[ ] Progress matches field data
[ ] Actuals match projections (within tolerance)

Verification Report Format

After running all phases, produce a verification report:

═══════════════════════════════════════════════════════════════
                    VERIFICATION REPORT
═══════════════════════════════════════════════════════════════

Output Type:    Cost Estimate
Project:        Downtown Office Tower
Generated:      2026-01-24 14:30:00
Verified By:    DDC Verification Loop v1.0

───────────────────────────────────────────────────────────────
PHASE 1: DATA INTEGRITY
───────────────────────────────────────────────────────────────
✓ Completeness      PASS    All required fields present
✓ Consistency       PASS    No inconsistencies found
✓ Referential       PASS    All references valid

───────────────────────────────────────────────────────────────
PHASE 2: BUSINESS LOGIC
───────────────────────────────────────────────────────────────
✓ Total Accuracy    PASS    Calculated: $1,523,456.78, Declared: $1,523,456.78
✓ Markup Check      PASS    All markups applied correctly
⚠ Range Check       WARN    3 items outside typical ranges

───────────────────────────────────────────────────────────────
PHASE 3: STANDARDS COMPLIANCE
───────────────────────────────────────────────────────────────
✓ CSI Codes         PASS    All codes valid
✓ CWICR Mapping     PASS    156/156 items mapped
✓ Unit Standards    PASS    All units metric

───────────────────────────────────────────────────────────────
PHASE 4: OUTPUT QUALITY
───────────────────────────────────────────────────────────────
✓ Excel Format      PASS    Headers, formatting correct
✓ Charts            PASS    All visualizations rendered
✓ Print Ready       PASS    Print area configured

───────────────────────────────────────────────────────────────
PHASE 5: CROSS-REFERENCE
───────────────────────────────────────────────────────────────
✓ vs BIM QTO        PASS    Variance: 0.2%
✓ vs Historical     PASS    Within expected range
⚠ vs Budget         WARN    5.3% over budget baseline

═══════════════════════════════════════════════════════════════
                         SUMMARY
═══════════════════════════════════════════════════════════════

Total Checks:       18
Passed:             16
Warnings:           2
Failed:             0

OVERALL STATUS:     ✓ READY FOR DELIVERY

Recommendations:
1. Review items outside typical ranges (see Appendix A)
2. Discuss budget variance with PM before submission

═══════════════════════════════════════════════════════════════

Automated Verification Pipeline

class ConstructionVerificationPipeline:
    """Automated verification for construction outputs"""

    def __init__(self, output_type: str):
        self.output_type = output_type
        self.phases = self._get_phases_for_type(output_type)

    def verify(self, output: dict, sources: list = None) -> VerificationReport:
        results = []

        for phase in self.phases:
            phase_result = phase.execute(output, sources)
            results.append(phase_result)

            # Stop on critical failure
            if phase_result.has_critical_failure():
                break

        return VerificationReport(
            output_type=self.output_type,
            phases=results,
            overall_status=self._calculate_overall_status(results)
        )

    def _calculate_overall_status(self, results: list) -> str:
        if any(r.has_failures() for r in results):
            return 'NOT READY - FIX REQUIRED'
        if any(r.has_warnings() for r in results):
            return 'READY WITH WARNINGS'
        return 'READY FOR DELIVERY'


# Usage
pipeline = ConstructionVerificationPipeline('cost_estimate')
report = pipeline.verify(estimate_output, sources=[bim_model, specifications])
print(report.to_markdown())

Integration with DDC Workflows

This verification skill integrates with other DDC skills:

Post-Estimation: Run after cost-estimation-* skills
Post-QTO: Run after qto-report skill
Post-Scheduling: Run after gantt-chart or 4d-simulation skills
Post-Pipeline: Run after etl-pipeline skill

Continuous Verification Mode

For long automation sessions, run verification at checkpoints:

Recommended checkpoints:
- After processing each BIM model
- After generating each major section of estimate
- After completing each phase of schedule
- Before any external data submission

Command: /verify-construction

Quality is not negotiable in construction. Verify before you deliver.

Other skills for the same job

different authors, same section of the catalogue

Doc Coauthoring

by anthropics

vendor ×10

Guide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting specs, or similar documentation tasks.

Intelligently organizes your files and folders across your computer by understanding context, finding duplicates, suggesting better structures, and automating cleanup tasks. Reduces cognitive load and keeps your digital workspace tidy without manual effort.

3k tokens

Domain Name Brainstormer

by frostant

×8

Generates creative domain name ideas for your project and checks availability across multiple TLDs (.com, .io, .dev, .ai, etc.). Saves hours of brainstorming and manual checking.

You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation.

Implements Manus-style file-based planning for complex tasks. Creates task_plan.md, findings.md, and progress.md. Use when starting complex multi-step tasks, research projects, or any task requiring >5 tool calls.

9k tokens scripts

Scientific Brainstorming

by christophacham

×3

Creative research ideation and exploration. Use for open-ended brainstorming sessions, exploring interdisciplinary connections, challenging assumptions, or identifying research gaps. Best for early-stage research planning when you do not have specific observations yet. For formulating testable hypotheses from data use hypothesis-generation.

5k tokens

GitHub Project Management

by ComeOnOliver

×3

Comprehensive GitHub project management with swarm-coordinated issue tracking, project board automation, and sprint planning

Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user wants to stress-test a plan, get grilled on their design, or mentions "grill me".

3k tokens

How to use it

Copy the folder

Take datadrivenconstruction/verification-loop-construction from the repository into ~/.claude/skills for personal use, or into .claude/skills inside a project.

Check the name does not clash

The agent identifies a skill by the name field in its header. Two skills with the same name cannot sit side by side — one of them will be ignored.