본문으로 건너뛰기
cccskills
무료GitHub에 공개

unified-deliverables-flow

Create spreadsheets, diagrams, and PDF reports in a staged workflow. Allocate execution attempts across deliverables, check generated files, and use simpler methods or alternative formats when tool errors or limited remaining attempts interrupt progress.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

원문원문 설명 보기

Generate spreadsheets, diagrams, and PDF reports in a phased workflow with explicit iteration budgets and error recovery

연동 도구Excel
설치 방법 보기

이럴 때 유용합니다

  • Preparing a comparison spreadsheet, diagram, and report
  • Allocating execution attempts across several deliverables
  • Recovering from file generation errors with fallback methods

포함된 파일(2)

  • SKILL.md14.9 KB
  • .skill_id45 B

SKILL.md(원문)

설치하기 전에 에이전트에게 전달되는 지시 내용을 확인할 수 있습니다.

Unified Multi-Deliverable Generation Workflow

This skill provides a structured, phased approach for creating multiple deliverable types (spreadsheets, diagrams, PDF reports) in a single cohesive workflow with explicit iteration budgeting to prevent premature exhaustion.

Overview

Use this workflow when you need to:

  • Generate multiple deliverable types (Excel, images, PDFs) in one task
  • Ensure balanced iteration allocation across all deliverables
  • Handle tool failures gracefully with retries and fallbacks
  • Produce downloadable artifacts with verified paths

Critical: Iteration Budget Allocation

Allocate iterations BEFORE starting work:

PhaseDeliverableBudgetCheckpoint
Phase 1Spreadsheet8 iterationsFile exists + readable
Phase 2Diagram8 iterationsFile exists + >1KB
Phase 3PDF Report8 iterationsFile exists + ARTIFACT_PATH output
BufferError recovery6 iterationsRemaining for retries
TotalAll deliverables30 iterationsAll verified

Rules:

  • Complete each phase before moving to the next
  • If a phase exceeds budget, use fallback strategy (see below)
  • Never spend >10 iterations on a single deliverable without checkpoint
  • Verify each deliverable before proceeding

Step-by-Step Instructions

Phase 0: Pre-Work Setup (1 iteration max)

  1. Define all deliverables explicitly:

    Deliverable 1: hardware_selection_table.xlsx (Excel with comparison data)
    Deliverable 2: cnc_workcell_layout.png (PNG diagram of layout)
    Deliverable 3: final_report.pdf (PDF summary report)
    
  2. Verify workspace is accessible:

    run_shell
      command: ls -la /workspace/ && mkdir -p /workspace/artifacts
    
  3. Set path variables for consistency:

    • All files go to /workspace/ or /workspace/artifacts/
    • Use absolute paths in all code
    • Never use relative paths like ./output.xlsx

Phase 1: Spreadsheet Generation (Budget: 8 iterations)

Step 1.1: Choose approach

MethodToolBest For
Direct Pythonexecute_code_sandboxQuick generation, simple tables
Shell agentshell_agentComplex logic, error recovery

Step 1.2: Create spreadsheet code

from openpyxl import Workbook
from openpyxl.styles import Font, Alignment, Border, Side

def create_hardware_table():
    wb = Workbook()
    ws = wb.active
    ws.title = "Hardware Selection"
    
    # Headers with styling
    headers = ["Component", "Option A", "Option B", "Option C", "Recommendation"]
    for col, header in enumerate(headers, 1):
        cell = ws.cell(row=1, column=col, value=header)
        cell.font = Font(bold=True)
        cell.alignment = Alignment(horizontal='center')
    
    # Data rows
    data = [
        ["Controller", "PLC-X100 ($500)", "PLC-Y200 ($650)", "PLC-Z300 ($800)", "PLC-Y200"],
        ["Motor", "Servo-500W ($300)", "Servo-750W ($400)", "Stepper-1kW ($250)", "Servo-750W"],
        ["Sensor", "Proximity-S1 ($50)", "Vision-V2 ($200)", "Laser-L3 ($150)", "Vision-V2"]
    ]
    
    for row_idx, row_data in enumerate(data, 2):
        for col_idx, value in enumerate(row_data, 1):
            ws.cell(row=row_idx, column=col_idx, value=value)
    
    # Auto-adjust column widths
    for col in ws.columns:
        max_length = max(len(str(cell.value)) for cell in col if cell.value)
        ws.column_dimensions[col[0].column_letter].width = min(max_length + 2, 25)
    
    output_path = '/workspace/hardware_selection_table.xlsx'
    wb.save(output_path)
    print(f'SUCCESS: Spreadsheet created at {output_path}')
    print(f'ARTIFACT_PATH:{output_path}')
    return output_path

create_hardware_table()

Step 1.3: Execute with verification

execute_code_sandbox
  code: <spreadsheet code from Step 1.2>
  language: python

Step 1.4: Verify (REQUIRED before proceeding)

run_shell
  command: ls -lh /workspace/*.xlsx && python3 -c "import openpyxl; openpyxl.load_workbook('/workspace/hardware_selection_table.xlsx')"

Checkpoint criteria:

  • ✓ File exists
  • ✓ Size > 1KB
  • ✓ No errors opening file
  • ✗ If failed, retry once with shell_agent, then proceed to Phase 2 with note

Phase 2: Diagram Generation (Budget: 8 iterations)

Step 2.1: Choose diagram type and tool

Diagram TypeToolLibrary
Flowchart/Blocksexecute_code_sandboxmatplotlib, graphviz
Layout/Floorplanshell_agentmatplotlib, PIL
Architectureexecute_code_sandboxmatplotlib, diagram

Step 2.2: Create diagram code (matplotlib example)

import matplotlib.pyplot as plt
import matplotlib.patches as patches

def create_layout_diagram():
    fig, ax = plt.subplots(figsize=(12, 8))
    ax.set_xlim(0, 100)
    ax.set_ylim(0, 80)
    ax.set_aspect('equal')
    ax.set_title('CNC Workcell Layout', fontsize=16, pad=20)
    
    # Define workcell components
    components = [
        {'label': 'CNC Machine', 'xy': (30, 40), 'w': 25, 'h': 20, 'color': '#3498db'},
        {'label': 'Robot Arm', 'xy': (60, 40), 'w': 15, 'h': 15, 'color': '#e74c3c'},
        {'label': 'Material Rack', 'xy': (10, 20), 'w': 15, 'h': 40, 'color': '#2ecc71'},
        {'label': 'Control Panel', 'xy': (75, 60), 'w': 12, 'h': 10, 'color': '#f39c12'},
        {'label': 'Safety Zone', 'xy': (25, 35), 'w': 50, 'h': 30, 'color': 'none', 'edge': '#95a5a6', 'dashed': True}
    ]
    
    for comp in components:
        if comp.get('edge'):
            rect = patches.Rectangle(
                (comp['xy'][0], comp['xy'][1]), 
                comp['w'], comp['h'],
                linewidth=2, 
                edgecolor=comp['edge'],
                facecolor=comp['color'],
                linestyle='--' if comp.get('dashed') else '-'
            )
        else:
            rect = patches.Rectangle(
                (comp['xy'][0], comp['xy'][1]), 
                comp['w'], comp['h'],
                linewidth=2, 
                edgecolor='black',
                facecolor=comp['color']
            )
        ax.add_patch(rect)
        ax.text(
            comp['xy'][0] + comp['w']/2, 
            comp['xy'][1] + comp['h']/2, 
            comp['label'],
            ha='center', va='center', fontsize=10, fontweight='bold'
        )
    
    plt.grid(True, alpha=0.3)
    plt.xlabel('Distance (m)')
    plt.ylabel('Distance (m)')
    
    output_path = '/workspace/cnc_workcell_layout.png'
    plt.savefig(output_path, dpi=150, bbox_inches='tight')
    plt.close()
    print(f'SUCCESS: Diagram created at {output_path}')
    print(f'ARTIFACT_PATH:{output_path}')
    return output_path

create_layout_diagram()

Step 2.3: Execute and verify

execute_code_sandbox
  code: <diagram code from Step 2.2>
  language: python

Verification:

run_shell
  command: ls -lh /workspace/*.png && file /workspace/*.png

Checkpoint criteria:

  • ✓ File exists
  • ✓ Size > 1KB
  • ✓ File type confirmed as PNG/image
  • ✗ If failed after 2 retries, use shell_agent fallback, then proceed

Phase 3: PDF Report Generation (Budget: 8 iterations)

Step 3.1: Prepare content aggregation

Gather data from previous phases:

  • Spreadsheet: key findings, recommendations
  • Diagram: visual summary
  • Additional: scoring, conclusions

Step 3.2: Create PDF generation code

from reportlab.lib import colors
from reportlab.lib.pagesizes import letter
from reportlab.platypus import SimpleDocTemplate, Paragraph, Spacer, Table, TableStyle, Image
from reportlab.lib.styles import getSampleStyleSheet, ParagraphStyle
from reportlab.lib.units import inch
from reportlab.lib.enums import TA_CENTER, TA_LEFT

def create_final_report(output_path):
    doc = SimpleDocTemplate(output_path, pagesize=letter,
                           rightMargin=0.75*inch, leftMargin=0.75*inch,
                           topMargin=0.75*inch, bottomMargin=0.75*inch)
    styles = getSampleStyleSheet()
    story = []
    
    # Custom styles
    title_style = ParagraphStyle('CustomTitle', parent=styles['Heading1'],
                                  fontSize=20, spaceAfter=30, alignment=TA_CENTER,
                                  fontName='Helvetica-Bold')
    heading_style = ParagraphStyle('CustomHeading', parent=styles['Heading2'],
                                    fontSize=14, spaceAfter=12, spaceBefore=20,
                                    fontName='Helvetica-Bold')
    
    # Title
    story.append(Paragraph("CNC Workcell Implementation Report", title_style))
    story.append(Spacer(1, 0.5*inch))
    
    # Executive Summary
    story.append(Paragraph("Executive Summary", heading_style))
    summary_text = """
    This report presents the hardware selection analysis and workcell layout 
    for the proposed CNC implementation. After evaluating multiple options across 
    controllers, motors, and sensors, recommended configurations balance cost, 
    performance, and reliability.
    """
    story.append(Paragraph(summary_text, styles['Normal']))
    story.append(Spacer(1, 0.3*inch))
    
    # Hardware Selection Table
    story.append(Paragraph("Hardware Selection Summary", heading_style))
    data = [
        ['Component', 'Recommended Option', 'Cost', 'Rationale'],
        ['Controller', 'PLC-Y200', '$650', 'Best cost/performance balance'],
        ['Motor', 'Servo-750W', '$400', 'Adequate power with precision'],
        ['Sensor', 'Vision-V2', '$200', 'Enables quality inspection']
    ]
    
    table = Table(data, colWidths=[1.5*inch, 1.7*inch, 1*inch, 2.3*inch])
    table.setStyle(TableStyle([
        ('BACKGROUND', (0, 0), (-1, 0), colors '#2c3e50'),
        ('TEXTCOLOR', (0, 0), (-1, 0), colors.whitesmoke),
        ('ALIGN', (0, 0), (-1, -1), 'CENTER'),
        ('FONTNAME', (0, 0), (-1, 0), 'Helvetica-Bold'),
        ('FONTSIZE', (0, 0), (-1, 0), 11),
        ('BOTTOMPADDING', (0, 0), (-1, 0), 12),
        ('BACKGROUND', (0, 1), (-1, -1), colors.beige),
        ('GRID', (0, 0), (-1, -1), 1, colors.black),
        ('VALIGN', (0, 0), (-1, -1), 'MIDDLE')
    ]))
    story.append(table)
    story.append(Spacer(1, 0.4*inch))
    
    # Layout Diagram Reference
    story.append(Paragraph("Workcell Layout", heading_style))
    story.append(Paragraph("See attached diagram: cnc_workcell_layout.png", styles['Normal']))
    story.append(Spacer(1, 0.2*inch))
    
    # Recommendations
    story.append(Paragraph("Implementation Recommendations", heading_style))
    rec_text = """
    1. Begin with controller installation and configuration
    2. Integrate motor drives and verify motion control
    3. Install and calibrate vision system
    4. Conduct safety zone validation
    5. Perform full system integration testing
    """
    story.append(Paragraph(rec_text, styles['Normal']))
    story.append(Spacer(1, 0.5*inch))
    
    # Footer
    story.append(Spacer(1, 0.5*inch))
    footer_style = ParagraphStyle('Footer', parent=styles['Normal'],
                                   fontSize=9, alignment=TA_CENTER, textColor=colors.grey)
    story.append(Paragraph("Generated by Unified Deliverables Workflow", footer_style))
    
    doc.build(story)
    print(f'SUCCESS: PDF report created at {output_path}')
    print(f'ARTIFACT_PATH:{output_path}')
    return output_path

create_final_report('/workspace/final_report.pdf')

Step 3.3: Execute with explicit artifact path

execute_code_sandbox
  code: <PDF code from Step 3.2>
  language: python

Step 3.4: Final verification

run_shell
  command: ls -lh /workspace/*.pdf && echo "---" && ls -lh /workspace/*.xlsx /workspace/*.png

Checkpoint criteria:

  • ✓ PDF file exists
  • ✓ Size > 10KB (reports should be substantial)
  • ✓ ARTIFACT_PATH was output
  • ✓ All three deliverables verified

Error Recovery & Fallback Strategies

When execute_code_sandbox fails repeatedly:

Strategy 1: Use shell_agent (more resilient)

shell_agent
  task: Create an Excel file at /workspace/hardware_selection_table.xlsx with hardware comparison data including controllers, motors, and sensors with costs and recommendations

Strategy 2: Simplify the code

  • Remove complex styling
  • Use basic libraries only (csv instead of openpyxl if needed)
  • Reduce dependencies

Strategy 3: Change output format

  • Excel → CSV if openpyxl fails
  • PNG → SVG if PIL/matplotlib fails
  • PDF → Markdown + convert later

When iteration budget is running low:

Iterations RemainingAction
< 10Skip non-critical formatting, use minimal viable output
< 5Use simplest possible implementation, skip verification
< 3Output text summary with file paths, request manual generation

Best Practices

  1. Phase sequentially - Complete and verify each phase before moving on
  2. Use absolute paths - Always /workspace/filename.ext, never relative
  3. Output ARTIFACT_PATH - Every successful generation must print this
  4. Verify before proceeding - Run shell check after each deliverable
  5. Track iteration count - Count tool calls, stop at 25 to leave buffer
  6. Simplify on failure - If complex code fails, strip to minimum viable
  7. Document decisions - Note why fallback was used for future reference

Quick Reference: Tool Selection

TaskPrimary ToolFallback Tool
Spreadsheetexecute_code_sandbox (openpyxl)shell_agent (pandas)
Diagramexecute_code_sandbox (matplotlib)shell_agent (graphviz)
PDF Reportexecute_code_sandbox (reportlab)shell_agent (fpdf)
Verificationrun_shell (ls, file)read_file (for content check)

Complete Workflow Checklist

  • Phase 0: Defined all 3 deliverables with paths
  • Phase 0: Verified workspace accessibility
  • Phase 1: Spreadsheet created and verified (< 8 iterations)
  • Phase 2: Diagram created and verified (< 8 iterations)
  • Phase 3: PDF created and verified (< 8 iterations)
  • All ARTIFACT_PATH outputs captured
  • Final verification: all files exist with appropriate sizes
  • Total iterations used ≤ 30

Troubleshooting

ProblemImmediate Action
execute_code_sandbox returns "[ERROR] unknown error"Retry once, then switch to shell_agent
File not found after executionCheck actual path with run_shell: ls /workspace/
ARTIFACT_PATH not outputRe-run with explicit print statement
Iteration budget nearly exhaustedSkip remaining deliverables, document what was created
Library import failsAdd !pip install <library> at code start

Related Tools

  • execute_code_sandbox - Primary tool for code execution (Python)
  • shell_agent - Fallback for complex tasks or when sandbox fails
  • run_shell - Verification and workspace inspection
  • read_file - Content verification (type: xlsx, png, pdf)
  • create_file - Alternative for text-based outputs

리뷰

아직 리뷰가 없습니다. 사용해 본 소감을 남겨 주세요.

댓글

로그인 없이 쓸 수 있습니다. 사용 후기, 질문, 작동하지 않은 환경 등을 편하게 남겨 주세요.

아직 댓글이 없습니다. 첫 댓글을 남겨 보세요.

보내면 바로 공개됩니다. 이메일 주소나 전화번호 같은 개인 정보는 쓰지 마세요. URL은 2개까지입니다.

0 / 1,000자

이 양식은 reCAPTCHA로 보호되며 Google 개인정보처리방침 및 서비스 약관이 적용됩니다.

같은 저장소의 스킬

Prepare audio stems, the separate parts of a track, for mixing by matching their durations with silence, loops, or crossfades. Verify each generated and processed file before moving to the next production step.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

  • Match audio parts with different durations
  • Extend drums and melodic parts using suitable methods
  • Check audio specifications before mixing
HKUDS/OpenSpace7.8천2026년 8월 13일 업데이트

Produce audio in separate stems, meaning individual instrument tracks, and verify each output as you go. Align track lengths with silence, loops, or crossfades, then apply effects, mix a master, and package the files.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

  • Align instrument tracks to a reference audio file
  • Choose how to extend tracks that are too short
  • Verify audio properties after generation and effects
HKUDS/OpenSpace7.8천2026년 8월 13일 업데이트

Build serverless endpoints that relay external API requests for a dashboard while keeping API keys on the server. Includes Vercel examples for input checks, caching, cross-origin access, and consistent JSON responses.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

  • Connect a dashboard to external APIs without exposing API keys
  • Create endpoints for stock quotes and news
  • Standardize API responses and error handling
HKUDS/OpenSpace7.8천2026년 8월 13일 업데이트

Produce a master track and separate audio parts through a structured Python workflow. Apply filters and reverb, export WAV files with consistent settings, package them in a ZIP archive, and check that the deliverables meet the requested specifications.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

  • Export a master track with separate audio parts
  • Apply filters and reverb to audio
  • Check duration and format across deliverables
HKUDS/OpenSpace7.8천2026년 8월 13일 업데이트

Keep document creation moving when web searches and page retrieval repeatedly fail. The skill preserves the original task, uses the model’s existing knowledge, and clearly identifies information that could not be checked against external sources.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

  • Continue drafting after repeated web retrieval failures
  • Preserve task requirements during network errors
  • Mark information that could not be externally verified
HKUDS/OpenSpace7.8천2026년 8월 13일 업데이트

When Python execution repeatedly fails in execute_code_sandbox, use a file-based alternative: save the script, run it through the shell, and capture results or errors. The guidance covers paths, imports, timeouts, structured output, and optional cleanup.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

  • Retry Python work through the shell after repeated sandbox failures
  • Run longer tasks with an adjusted timeout
  • Execute Python work that needs multiple files or system utilities
HKUDS/OpenSpace7.8천2026년 8월 13일 업데이트

HKUDS의 스킬 모두 보기

이 스킬의 문제 신고