This guide provides step-by-step instructions for accomplishing common tasks using Power Tools in elvex. Each section addresses a specific goal with clear, actionable steps.
Creating Documents
How to Create a PDF Report with Charts
Goal: Generate a professional PDF document that includes data visualizations.
When to use: When you need a polished, shareable report with embedded graphics.
Steps:
Prepare your data: Have your data ready (as a CSV file, or as specific values to share with the agent).
Request the report: Be specific about what you want:
Create a PDF report analyzing Q1 sales performance. Include: - A bar chart showing monthly revenue - A line chart showing customer growth - A summary section with key insights Use professional styling and make it print-ready.
Provide data: Either:
Upload a data file, or
Include specific numbers in your request
Wait for generation: The agent will:
Read the markdown-to-pdf skill
Create visualizations using matplotlib or seaborn
Generate the PDF with embedded charts
Save it to the outputs directory
Review and download: Click the provided link to view or download your PDF.
Iterate if needed: Request changes like "Make the charts bigger" or "Add a table summarizing the data."
Tips:
Specify PDF explicitly—don't just say "report"
Mention any specific styling requirements (colors, fonts, etc)
If you have a brand style guide, describe it or upload an example
For complex reports, break into sections and build iteratively
Example requests:
"Create a 5-page PDF report on project status with timeline charts"
"Generate a PDF invoice with itemized charges and company logo"
"Make a PDF white paper about AI trends with embedded research charts"
How to Create an Excel Spreadsheet with Formulas
Goal: Generate an Excel file with formatted data, formulas, and multiple sheets.
When to use: When you need structured data with calculations, or templates for data entry.
Steps:
Define your requirements:
Create an Excel spreadsheet for tracking project expenses with: - Sheet 1: Monthly breakdown with category columns - Sheet 2: Summary with total by category - Include formulas to calculate totals automatically - Format headers and use appropriate number formats for currency
The agent will:
Read the xlsx skill for Excel best practices
Create the workbook with openpyxl
Add data, formulas, and formatting
Create multiple sheets if requested
Save to outputs directory
Download and test: Open in Excel or Google Sheets to verify formulas work.
Refine if needed: "Add conditional formatting to highlight expenses over $1000" or "Create a chart in Sheet 3 showing expense trends."
Tips:
Specify sheet names and structure clearly
Mention specific Excel features you want (charts, pivot tables, conditional formatting)
Provide sample data or structures if you have specific layouts in mind
Test formulas to ensure they work as expected
Example requests:
"Create an Excel budget template with expense categories and formula-driven totals"
"Generate a timesheet spreadsheet with automatic hourly rate calculations"
"Make an Excel inventory tracker with low stock alerts using conditional formatting"
How to Create a PowerPoint Presentation
Goal: Build a presentation with multiple slides, charts, and professional design.
When to use: For pitch decks, training materials, or any slide-based content.
Steps:
Outline your presentation:
Create a PowerPoint about our product launch with these slides: 1. Title slide with company logo 2. Problem statement 3. Our solution (with feature screenshots) 4. Market size (with pie chart) 5. Financial projections (with bar chart) 6. Call to action Use a clean, professional theme.
Provide assets: Upload any images, logos, or data files you want included.
The agent will:
Read the pptx skill
Create the presentation structure
Add content to each slide
Include requested charts or images
Apply formatting and layout
Review the presentation: Download and open in PowerPoint or Google Slides.
Iterate: "Add speaker notes to each slide" or "Change the color scheme to match our brand colors (blue and orange)."
Tips:
Specify the number of slides and what goes on each
Mention any specific themes, colors, or styles
Upload logos, images, or brand assets before requesting
For charts, describe the data or upload data files
Consider requesting speaker notes for complex presentations
Example requests:
"Create a 10-slide pitch deck for investors with financial charts"
"Generate a training presentation about our new software feature"
"Make a quarterly review presentation with performance charts for each department"
How to Create a Word Document
Goal: Generate a formatted Word document with sections, styles, and embedded elements.
When to use: For letters, proposals, contracts, or any formal document.
Steps:
Describe the document:
Create a Word document business proposal with: - Cover page with title and date - Executive summary (1 page) - Detailed scope of work (2 pages) - Pricing table - Terms and conditions Use professional formatting with headers and footers.
The agent will:
Read the docx skill
Create the document structure
Add content with proper formatting
Include tables, lists, or other elements
Apply styles for headers, body text, etc.
Download and review: Open in Microsoft Word or compatible software.
Refine: "Add a table of contents" or "Include page numbers in the footer."
Tips:
Specify sections and their approximate length
Mention any specific formatting requirements
Upload any content that should be included (terms and conditions, etc)
Request headers, footers, or page numbering if needed
For contracts or formal documents, be very explicit about structure
Example requests:
"Create a Word document cover letter for a job application"
"Generate a proposal template for consulting services"
"Make a project plan document with sections for timeline, deliverables, and team structure"
Working with Data
How to Analyze a CSV File and Create Visualizations
Goal: Process a CSV file to extract insights and create visual representations.
When to use: When you have data in CSV format and need analysis or charts.
Steps:
Upload your CSV: Use the attachment button to upload your data file.
Request analysis:
Analyze this sales data CSV and: - Calculate total revenue, average order value, and customer count - Identify top 10 products by revenue - Create a time series chart showing daily revenue - Generate a bar chart of revenue by product category - Provide a summary of key findings Save all charts and create a PDF report with the analysis.
The agent will:
Load the CSV with pandas
Perform requested calculations
Create visualizations with matplotlib/seaborn
Generate summary statistics
Compile into a report
Review the results: You'll receive charts and/or a compiled report.
Dig deeper: "What's the correlation between price and sales volume?" or "Show me a monthly breakdown for the top 5 products."
Tips:
Preview your CSV first to understand the structure
Be specific about which columns contain what data
Mention any data cleaning needed (missing values, date formats, etc)
Request multiple visualization types to see different perspectives
Ask for specific statistical measures if you know what you need
Example requests:
"Analyze this customer data CSV and create demographic charts"
"Process this financial data and calculate year-over-year growth"
"Load this survey data and create visualizations showing response distributions"
How to Convert Data Between Formats
Goal: Transform data from one format to another (e.g., CSV to JSON, Excel to CSV).
When to use: When you need data in a different format for another tool or system.
Steps:
Upload the source file: Attach the file you want to convert.
Specify the conversion:
Convert this Excel file to CSV format. Keep only the "Sales Data" sheet and exclude the summary sheet.
Or:
Convert this CSV to JSON format. Each row should be an object with column names as keys.
The agent will:
Read the source file
Parse the data
Transform to the target format
Handle any necessary restructuring
Save the converted file
Download the result: You'll get a link to the converted file.
Validate: Open in the target application to verify correctness.
Tips:
Specify which sheet to convert if working with Excel files
Mention any filtering or subsetting needed
Describe the desired JSON structure if converting to JSON
For CSV, clarify delimiter preferences (comma, tab, etc)
Test the converted file in your target system
Example requests:
"Convert this JSON API response to a CSV file"
"Transform this XML file into an Excel spreadsheet with proper columns"
"Take this multi-sheet Excel workbook and create separate CSV files for each sheet"
How to Merge Multiple Data Files
Goal: Combine data from several sources into a single file.
When to use: When you have related data in separate files that needs to be consolidated.
Steps:
Upload all source files: Attach each file that should be merged.
Describe the merge logic:
Merge these three CSV files (Jan, Feb, Mar sales data) into a single file: - Stack them vertically (all have the same columns) - Add a "Month" column to identify the source - Sort by date - Remove any duplicate entries Save as both CSV and Excel.
The agent will:
Load all source files
Perform the merge with specified logic
Handle any data cleaning
Create the consolidated output
Review the merged file: Check that all data is present and correctly combined.
Adjust if needed: "The date columns don't match format—standardize them before merging."
Tips:
Clearly explain the relationship between files
Specify how to handle conflicts or duplicates
Mention any column mapping needed (if columns have different names)
Describe desired sorting or organization in the output
Request validation checks (row counts, unique IDs, etc)
Example requests:
"Merge these department expense reports into one master report"
"Combine customer data from these three regional files, matching by email address"
"Join this sales data with this customer data using the customer ID column"
How to Clean and Transform Data
Goal: Prepare messy data for analysis by handling missing values, fixing formats, etc.
When to use: When your data has issues that prevent analysis or use.
Steps:
Upload the raw data: Attach the file that needs cleaning.
Describe the problems and desired cleaning:
Clean this dataset: - Remove rows where Revenue is null or zero - Standardize the Date column to YYYY-MM-DD format - Strip whitespace from product names - Convert price columns to numeric (remove $ symbols) - Fill missing Region values with "Unknown" - Remove duplicate rows based on Order ID Provide a summary of changes made.
The agent will:
Load and examine the data
Apply each cleaning step
Track and report changes
Save the cleaned data
Review the summary: Check what was changed and verify it matches your intent.
Further refinement: "Also uppercase all product categories" or "Cap outlier values above $10,000."
Tips:
Be specific about what defines "clean" for your use case
Mention which columns are most important to get right
Request a before/after summary to understand changes
Ask for removed rows to be saved separately if you want to review them
Test the cleaned data with a sample analysis to verify quality
Example requests:
"Clean this survey data by removing incomplete responses and standardizing ratings"
"Fix date formats and handle missing values in this time series data"
"Normalize this customer database by standardizing addresses and phone numbers"
Creating and Processing Files
How to Generate Scripts
Goal: Create working script files for specific purposes.
When to use: When you need scripts, configuration files, or snippets for a project.
Steps:
Describe what the script should do:
Create a Python script that: - Reads CSV files from an input directory - Calculates summary statistics for numeric columns - Generates a report PDF for each file - Saves reports to an output directory Include error handling and logging.
Specify requirements:
Programming language
Any libraries or frameworks to use
Expected input/output format
Error handling needs
The agent will:
Write the script following best practices
Add comments explaining key sections
Create necessary helper functions
Save to the outputs directory
Download and test: Run the script in your environment to verify it works.
Iterate: "Add command-line arguments for the input directory" or "Include a progress bar for large files."
Tips:
Specify the language and any version requirements
Mention libraries you prefer or restrictions (e.g., "use only standard library")
Describe edge cases or error conditions to handle
Request comments and documentation
Ask for example usage or test cases
Example requests:
"Create a bash script to backup files to S3 with timestamp names"
"Generate a JavaScript function to validate email addresses with regex"
"Write a Python class for managing database connections with connection pooling"
How to Extract Content from Documents
Goal: Pull specific information from PDFs, Word docs, or other files.
When to use: When you need data from documents that aren't easily copy-paste accessible.
Steps:
Upload the document: Attach the PDF, Word file, or other document.
Specify what to extract:
Extract all invoice data from this PDF: - Invoice number - Date - Line items (product, quantity, price) - Total amount Save as a structured CSV file.
The agent will:
Use appropriate tools to read the document (PyPDF2, python-docx, etc)
Extract the requested content
Structure it as specified
Handle any format conversion
Review the extracted data: Verify accuracy and completeness.
Refine: "Also extract the customer name and address" or "Format the date as MM/DD/YYYY."
Tips:
For PDFs, understand that complex layouts may be challenging
Specify desired output format (CSV, JSON, text, etc)
Mention any patterns or sections to focus on
For images in PDFs, note that OCR capabilities may be limited
Test with a small sample first if processing many documents
Example requests:
"Extract all email addresses from this PDF document"
"Pull the executive summary section from this Word document"
"Get all tables from this PDF and save each as a separate CSV"
How to Process Images
Goal: Manipulate, analyze, or extract information from images.
When to use: When you need to resize, crop, convert, or analyze images.
Steps:
Upload the image(s): Attach the image files.
Describe the processing:
Process this image: - Resize to 800x600 pixels - Convert to grayscale - Enhance contrast - Save as both PNG and JPEG
Or:
Analyze this chart image and extract the data values shown in the bar chart.
The agent will:
Use PIL/Pillow or similar libraries
Apply requested transformations
Perform analysis if requested
Save processed images
Download and verify: Check that processing matches expectations.
Tips:
Specify exact dimensions for resizing
Mention desired output formats
For analysis tasks, note that the agent can see images and describe them
For OCR or text extraction, results depend on image quality
Batch processing is possible—upload multiple images and describe the operation once
Example requests:
"Create thumbnails of these product images (200x200) for a website"
"Convert this PNG logo to SVG format with transparent background"
"Crop this screenshot to show only the chart area"
Advanced Workflows
How to Create a Multi-Step Data Pipeline
Goal: Process data through several transformation steps to produce final outputs.
When to use: For complex analysis requiring multiple stages of processing.
Steps:
Upload source data: Provide all input files.
Describe the pipeline:
Build a data pipeline: 1. Load sales data from CSV 2. Clean: remove duplicates, handle missing values 3. Enrich: add product category from products.csv lookup 4. Calculate: daily, weekly, monthly aggregates 5. Visualize: create 5 charts showing different trends 6. Report: compile into PDF with charts and summary tables Save intermediate files at each step for review.
The agent will:
Execute each step sequentially
Save intermediate outputs
Create visualizations
Compile final deliverable
Review intermediate steps: Check outputs at each stage to validate processing.
Adjust if needed: "Recalculate step 4 using median instead of mean" or "Add another chart type in step 5."
Tips:
Number the steps clearly
Specify what should be saved at each stage
Mention any validation checks between steps
For long pipelines, consider breaking into multiple conversations
Ask for status updates if processing takes time
Example requests:
"Build a pipeline: extract data from API → clean → analyze → visualize → create dashboard"
"Process survey responses through validation → categorization → statistical analysis → report generation"
"ETL pipeline: extract from Excel → transform formats → calculate metrics → load into summary file"
Conclusion
This guide has covered the most common tasks you'll accomplish with Power Tools. As you gain experience, you'll discover even more possibilities and develop your own workflows.
Remember:
Be specific in your requests
Trust the agent's technical approach
Iterate to refine outputs
Learn from each interaction
Power Tools puts sophisticated computing capabilities at your fingertips through natural language. With practice, you'll be creating complex outputs in minutes that would take hours manually.