datadrivenconstruction/ddc_skills_for_ai_agents_in_construction/daily-progress-report

Claude Code and agent skill by datadrivenconstruction/ddc_skills_for_ai_agents_in_construction

Trust tier 2
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Generate automated daily progress reports from site data. Track work completed, labor hours, equipment usage, and weather conditions.

Install

npx vskill@latest install datadrivenconstruction/ddc_skills_for_ai_agents_in_construction/daily-progress-report

SKILL.md

Daily Progress Report Generator

Business Case

Problem Statement

Site managers spend hours creating daily reports:

  • Manual data collection
  • Inconsistent formats
  • Delayed submissions
  • Missing information
Solution

Automated daily progress report generation from structured site data inputs.

Technical Implementation

python
import pandas as pd
from datetime import datetime, date
from typing import Dict, Any, List
from dataclasses import dataclass
from enum import Enum
class WeatherCondition(Enum):
CLEAR = "clear"
CLOUDY = "cloudy"
RAIN = "rain"
SNOW = "snow"
WIND = "wind"
EXTREME = "extreme"
class WorkStatus(Enum):
COMPLETED = "completed"
IN_PROGRESS = "in_progress"
DELAYED = "delayed"
NOT_STARTED = "not_started"
@dataclass
class WorkActivity:
activity_id: str
description: str
location: str
planned_qty: float
actual_qty: float
unit: str
status: WorkStatus
crew_size: int
hours_worked: float
notes: str = ""
@dataclass
class LaborEntry:
trade: str
company: str
workers: int
hours: float
overtime_hours: float = 0
@dataclass
class EquipmentEntry:
equipment_type: str
equipment_id: str
hours_used: float
status: str # active, idle, maintenance
operator: str = ""
@dataclass
class DailyReport:
report_date: date
project_name: str
project_number: str
weather: WeatherCondition
temperature_high: float
temperature_low: float
work_activities: List[WorkActivity]
labor: List[LaborEntry]
equipment: List[EquipmentEntry]
delays: List[str]
safety_incidents: int
visitors: List[str]
deliveries: List[str]
prepared_by: str
class DailyProgressReporter:
"""Generate daily progress reports."""
def __init__(self, project_name: str, project_number: str):
self.project_name = project_name
self.project_number = project_number
def create_report(self,
report_date: date,
weather: WeatherCondition,
temp_high: float,
temp_low: float,
prepared_by: str) -> DailyReport:
"""Create new daily report."""
return DailyReport(
report_date=report_date,
project_name=self.project_name,
project_number=self.project_number,
weather=weather,
temperature_high=temp_high,
temperature_low=temp_low,
work_activities=[],
labor=[],
equipment=[],
delays=[],
safety_incidents=0,
visitors=[],
deliveries=[],
prepared_by=prepared_by
)
def add_work_activity(self,
report: DailyReport,
activity_id: str,
description: str,
location: str,
planned_qty: float,
actual_qty: float,
unit: str,
crew_size: int,
hours_worked: float,
notes: str = ""):
"""Add work activity to report."""
# Determine status
if actual_qty >= planned_qty:
status = WorkStatus.COMPLETED
elif actual_qty > 0:
status = WorkStatus.IN_PROGRESS
elif actual_qty == 0 and planned_qty > 0:
status = WorkStatus.DELAYED
else:
status = WorkStatus.NOT_STARTED
activity = WorkActivity(
activity_id=activity_id,
description=description,
location=location,
planned_qty=planned_qty,
actual_qty=actual_qty,
unit=unit,
status=status,
crew_size=crew_size,
hours_worked=hours_worked,
notes=notes
)
report.work_activities.append(activity)
def add_labor(self,
report: DailyReport,
trade: str,
company: str,
workers: int,
hours: float,
overtime_hours: float = 0):
"""Add labor entry."""
report.labor.append(LaborEntry(
trade=trade,
company=company,
workers=workers,
hours=hours,
overtime_hours=overtime_hours
))
def add_equipment(self,
report: DailyReport,
equipment_type: str,
equipment_id: str,
hours_used: float,
status: str,
operator: str = ""):
"""Add equipment entry."""
report.equipment.append(EquipmentEntry(
equipment_type=equipment_type,
equipment_id=equipment_id,
hours_used=hours_used,
status=status,
operator=operator
))
def calculate_summary(self, report: DailyReport) -> Dict[str, Any]:
"""Calculate report summary metrics."""
total_workers = sum(l.workers for l in report.labor)
total_manhours = sum(l.workers * l.hours for l in report.labor)
total_overtime = sum(l.workers * l.overtime_hours for l in report.labor)
equipment_hours = sum(e.hours_used for e in report.equipment)
completed = sum(1 for a in report.work_activities if a.status == WorkStatus.COMPLETED)
in_progress = sum(1 for a in report.work_activities if a.status == WorkStatus.IN_PROGRESS)
delayed = sum(1 for a in report.work_activities if a.status == WorkStatus.DELAYED)
return {
'total_workers': total_workers,
'total_manhours': round(total_manhours, 1),
'total_overtime': round(total_overtime, 1),
'equipment_hours': round(equipment_hours, 1),
'activities_completed': completed,
'activities_in_progress': in_progress,
'activities_delayed': delayed,
'safety_incidents': report.safety_incidents,
'deliveries_count': len(report.deliveries)
}
def export_to_excel(self, report: DailyReport, output_path: str) -> str:
"""Export report to Excel."""
with pd.ExcelWriter(output_path, engine='openpyxl') as writer:
# Header
header_df = pd.DataFrame([{
'Project': report.project_name,
'Project #': report.project_number,
'Date': report.report_date,
'Weather': report.weather.value,
'High Temp': report.temperature_high,
'Low Temp': report.temperature_low,
'Prepared By': report.prepared_by
}])
header_df.to_excel(writer, sheet_name='Summary', index=False)
# Work Activities
if report.work_activities:
activities_df = pd.DataFrame([
{
'Activity ID': a.activity_id,
'Description': a.description,
'Location': a.location,
'Planned': a.planned_qty,
'Actual': a.actual_qty,
'Unit': a.unit,
'Status': a.status.value,
'Crew': a.crew_size,
'Hours': a.hours_worked,
'Notes': a.notes
}
for a in report.work_activities
])
activities_df.to_excel(writer, sheet_name='Work Activities', index=False)
# Labor
if report.labor:
labor_df = pd.DataFrame([
{
'Trade': l.trade,
'Company': l.company,
'Workers': l.workers,
'Hours': l.hours,
'Overtime': l.overtime_hours,
'Total Hours': l.workers * (l.hours + l.overtime_hours)
}
for l in report.labor
])
labor_df.to_excel(writer, sheet_name='Labor', index=False)
# Equipment
if report.equipment:
equip_df = pd.DataFrame([
{
'Type': e.equipment_type,
'ID': e.equipment_id,
'Hours': e.hours_used,
'Status': e.status,
'Operator': e.operator
}
for e in report.equipment
])
equip_df.to_excel(writer, sheet_name='Equipment', index=False)
return output_path
def generate_text_report(self, report: DailyReport) -> str:
"""Generate text version of report."""
summary = self.calculate_summary(report)
lines = [
f"DAILY PROGRESS REPORT",
f"=" * 50,
f"Project: {report.project_name}",
f"Project #: {report.project_number}",
f"Date: {report.report_date}",
f"Prepared by: {report.prepared_by}",
f"",
f"WEATHER CONDITIONS",
f"-" * 30,
f"Conditions: {report.weather.value}",
f"Temperature: {report.temperature_low}°C - {report.temperature_high}°C",
f"",
f"SUMMARY",
f"-" * 30,
f"Total Workers: {summary['total_workers']}",
f"Total Man-hours: {summary['total_manhours']}",
f"Equipment Hours: {summary['equipment_hours']}",
f"Activities Completed: {summary['activities_completed']}",
f"Activities In Progress: {summary['activities_in_progress']}",
f"Activities Delayed: {summary['activities_delayed']}",
f"Safety Incidents: {summary['safety_incidents']}",
]
if report.delays:
lines.extend([f"", f"DELAYS", f"-" * 30])
for delay in report.delays:
lines.append(f"• {delay}")
return "\n".join(lines)

Quick Start

python
from datetime import date
# Initialize reporter
reporter = DailyProgressReporter("Office Tower A", "PRJ-2024-001")
# Create report
report = reporter.create_report(
report_date=date.today(),
weather=WeatherCondition.CLEAR,
temp_high=28,
temp_low=18,
prepared_by="John Smith"
)
# Add activities
reporter.add_work_activity(
report,
activity_id="A-101",
description="Pour concrete slab Level 3",
location="Level 3, Zone A",
planned_qty=150,
actual_qty=150,
unit="m3",
crew_size=8,
hours_worked=10
)
# Add labor
reporter.add_labor(report, "Concrete", "ABC Concrete Co", 8, 10, 2)
# Export
reporter.export_to_excel(report, "daily_report.xlsx")

Common Use Cases

1. Generate Text Summary
python
text = reporter.generate_text_report(report)
print(text)
2. Track Delays
python
report.delays.append("Weather delay - rain from 14:00-16:00")
report.delays.append("Material delivery late by 2 hours")
3. Calculate Metrics
python
summary = reporter.calculate_summary(report)
print(f"Productivity: {summary['total_manhours']} man-hours")

Resources

  • DDC Book: Chapter 4.1 - Site Data Collection

Security

  • Security scanPassed, 100 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

Repository stars
337
Repository forks
83
7-day trend score
29

Stars and forks are for the whole GitHub repository, not this skill alone. Refreshed 7 May 2026, 5 months ago.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/datadrivenconstruction/ddc_skills_for_ai_agents_in_construction/daily-progress-report/badge)](https://verified-skill.com/skills/datadrivenconstruction/ddc_skills_for_ai_agents_in_construction/daily-progress-report)

Something wrong with this skill or its page? Tell us.