project-data-intel

name	project-data-intel
description	Comprehensive Project Data Intelligence dashboard — the single source of truth for ALL extracted project data. Template-based architecture with 40+ sections covering building data, quantities, financials, specifications, project controls, spatial analysis, and AI rendering. Triggers: "/data", "show me all the data", "project intelligence", "room browser", "quantity dashboard", "material calculator", "bid comparison", "project brain dashboard", "show everything we know", "data dashboard", "what data do we have".
version	3.0.0

Project Data Intelligence (v3.0) — SKILL Documentation

Overview

Comprehensive extraction and intelligence system for construction project documents.

1. Philosophy: "No Data Left Behind"

The project-data-intel skill is the single source of truth for all extracted construction project data. It operates on a universal principle: every piece of information captured during project setup, document processing, and field operations should be searchable, viewable, and actionable from a single, role-aware interface.

Key design principles:

• Template-based architecture: Static HTML template ships with the plugin, dynamic data is injected
• Project-agnostic: Works for any building type (senior care, office, industrial, healthcare, etc.)
• Incremental updates: Only changed data is re-serialized; unchanged sections remain cached
• Role-aware rendering: Superintendent, PM, Owner, and Architect each see customized views
• AI-native: Chatbot and rendering system leverage Anthropic API for context-aware insights
• Offline-capable: Keyword search fallback when API unavailable

---

2. Architecture: Template + Data Injection

The v3.0 dashboard uses a separation of concerns model:

File Structure

AI - Project Brain/
├── {PROJECT_CODE}_Data_Intel.html        ← Static template (10,000+ lines, generated once)
├── {PROJECT_CODE}_data.js                ← Dynamic data payload (regenerated each run)
├── references/
│   └── data-intel-template.html           ← Master template in plugin (never regenerated)
└── logs/
    └── data-intel-generation.log          ← Audit trail

Template Lifecycle

1. First run: Copy data-intel-template.html from plugin → {PROJECT_CODE}_Data_Intel.html
2. Subsequent runs: Never touch the HTML file; only regenerate {PROJECT_CODE}_data.js
3. Rationale: Template is stable (~10,000 lines), data evolves rapidly. Decoupled updates reduce file size and generation time.

Data Payload Structure (data.js)

const PROJECT_DATA = {
  // Core project metadata and configuration
  config: {
    projectCode, projectName, address, buildingType, occupancy,
    company, startDate, endDate, architect, pm, super, status
  },

  // Spatial & geometric data
  spatial: {
    gridLines: { x: [...], y: [...] },
    buildingAreas: { footprint, perimeter, floors, zones },
    rooms: [{id, name, area, type, occupancy, finishes, doors, equipment}],
    floorPlans: [{floor, gridRef, area, roomCount}]
  },

  // Building specifications and systems
  specs: {
    building: {structure, foundation, roof, exterior, interior},
    pemb: {supplier, model, dimensions, frames, reactions, design_loads},
    cfs: {profile, gauge, spacing, height},
    concrete: {bearing, sog_interior, sog_exterior, rebar, mixes, anchor_bolts, testing},
    mep: {
      equipment: [
        {
          id: "RTU-1", tag: "RTU-1", type: "Rooftop Unit",
          discipline: "mechanical",    // mechanical|plumbing|electrical|fire_protection
          system: "hvac",              // hvac|exhaust|plumbing|electrical_power|lighting|fire
          description: "10-Ton Rooftop Unit",
          location: { grid: "C-D/3-4", room: null, mounting: "Roof" },
          : { : , : , :  },
          : { : , : , : , :  },
          : [,,],
          : , : ,
          : , : 
        }
      ],
      : [{
        : , : ,
        : , : , : ,
        : , : ,
        : , : , : ,
        : [{: , : , : , : , : , : }],
        : {: , : , : , : }
      }],
      : {
        : {: , : , : },
        : {: , : },
        : [], : [],
        : , : , : 
      },
      : {
        : [{: , : , : , : , : , : , : }],
        : [{: , : , : , : , : , : , : , : }]
      },
      : {
        : [{: , : , : , : }],
        : [{: , : , : }]
      },
      : {
        : , : , : ,
        : {: , : }, : {: , : },
        : [{: , : , : , : }],
        : 
      },
      : [{: , : , : , : , : }],
      : [],
      : {
        : [],
        : ,
        : 
      }
    },
    : {flooring, wall_finishes, ceiling, paint, acoustical},
    : [{opening, type, material, hardware, schedule, supplier}],
    : [{opening, type, material, schedule, supplier}],
    : [{id, location, type, material, supplier}]
  },

  
  : {
    : [{task, duration, start, finish, preds, slack, critical}],
    : [{event, plannedDate, actualDate, status}],
    : [],
    : {},
    : [{item, supplier, orderDate, deliveryDate, duration}]
  },

  
  : {
    : [{name, role, phone, email, address}],
    : [{name, company, phone, email}],
    : [{name, company, phone, email}],
    : [{discipline, name, company, phone, email}],
    : [{trade, company, name, phone, email, contract}],
    : [{category, name, contact, phone, email, terms}],
    : [{agency, contact, phone, email}],
    : [{equipment, name, phone, license}]
  },

  
  : {
    : {pending, review, approved, rejected, deferred},
    : [{id, item, supplier, submitted, days_in_review, status, approver, notes}]
  },

  : {
    : [{po_number, supplier, item, quantity, unit_price, total, date, status}],
    : [{item, po, location, delivery_date, received, notes}],
    : [{item, : [{name, price, delivery, lead_time}]}],
    : [{priority, item, issue, action_required, due_date}]
  },

  
  : {
    : [{hp_id, description, trigger, responsible, status, date_completed}],
    : [{test_id, type, location, result, pass_fail, engineer, date}],
    : [{agency, type, scope, inspection_date, result, notes}],
    : [{id, location, date, category, photographer, notes}]
  },

  
  : [{date, superintendent, weather, temp, crew_size, work_summary, issues, safety_events}],

  
  : {
    : [{rfi_id, date_submitted, question, submitted_by, assigned_to, days_open, response, date_resolved}],
    : {design, constructability, materials, coordination, other}
  },

  
  : [{date, type, attendees, agenda, decisions, action_items}],

  
  : [{co_id, description, cost_impact, schedule_impact, status, date, approvals}],

  
  : [{id, event, cause, trade, duration_days, impact, mitigation, status}],

  
  : {
    : {total_contract, total_completed, percent_complete, total_invoiced, retainage},
    : [{period, contractor, amount, completed_value, retainage, approved_date}]
  },

  
  : [{id, location, item, trade, priority, assigned_to, status, due_date}],

  
  : {
    : [{date, location, category, photographer, file_reference}],
    : [{date, phase, photographer, file_reference}],
    : [{drawing, date, marked_up_by, file_reference}]
  },

  
  : [{prompt, type, model, generated_date, ai_image_reference}]
};

  = {
  : process.. || ,
  : process.. || ,
  : process.. || 
};

  = ; 

  = {
  : ,
  : ,
  : ,
  : {
    : {: , : , : },
    : {: , : , : },
    : {: , : , : },
    
  },
  : {
    : ,
    : ,
    : ,
    : ,
    : ,
    : ,
    : ,
    : ,
    : ,
    
  }
};

---

3. Incremental Update System

The dashboard uses smart regeneration to minimize file I/O and improve performance:

Update Algorithm

function updateProjectData(projectPath, sourceFiles) {
  // Step 1: Check if data.js exists
  if (NOT EXISTS data.js) {
    // Full generation path
    generateAllJSON(sourceFiles) → data.js
    copyTemplate(plugin/references/data-intel-template.html) → {PROJECT_CODE}_Data_Intel.html
    return "NEW_DASHBOARD_CREATED"
  }

  // Step 2: Load existing manifest
  manifest = parseManifest(data.js)

  // Step 3: Compute MD5 for each source
  for each sourceFile in [config, spatial, specs, schedule, directory, ...]:
    currentHash = MD5(sourceFile)
    storedHash = manifest.files[sourceFile.name].md5
    if (currentHash !== storedHash) {
      CHANGED_FILES.push(sourceFile)
    }

  // Step 4: Selective serialization
  if (CHANGED_FILES.length === 0) {
    return "NO_CHANGES_DETECTED"
  }

  // Step 5: Regenerate only changed sections
  for each changedFile in CHANGED_FILES:
    re_parse(changedFile) → update PROJECT_DATA[section]
    update manifest.files[changedFile.name].md5
    log("Updated: " + changedFile.name)

  // Step 6: Write data.js
  serializeToJS(PROJECT_DATA, API_KEYS, ACTIVE_ROLE, manifest) → data.js
  updateGenerationLog()
  return "UPDATED: " + CHANGED_FILES.length + " sections"
}

Fallback: Corrupted Manifest

If DATA_MANIFEST is invalid, the system falls back to full regeneration with validation:

1. Parse all JSON source files fresh
2. Validate schema against PROJECT_DATA structure
3. Recompute all MD5 hashes
4. Regenerate data.js
5. Log warning in generation log

---

---

Extended reference: Detailed examples, templates, scoring rubrics, and best practices are in references/skill-detail.md.