evolink-data-analysis（evolink 数据分析）

Data Analysis Assistant

AI-powered data analysis with decision-first methodology and statistical rigor.

Powered by Evolink.ai

When to Use

Use this skill when the user needs to:

• Analyze data from CSV, Excel, JSON files
• Find patterns, trends, or anomalies
• Understand metrics and KPIs
• Test hypotheses or A/B tests
• Perform cohort or funnel analysis
• Debug data quality issues
• Generate insights for decision-making

Core Principle: Analysis without a decision is just arithmetic. Always clarify what would change if the analysis shows X vs Y.

Usage

{baseDir}/scripts/analyze.sh <file_path> "<analysis_question>"

Configuration

👉 Get free API key

Example

bash scripts/analyze.sh sales_data.csv "What are the top 3 revenue drivers this quarter?"

Output:

📊 Analyzing: sales_data.csv
❓ Question: What are the top 3 revenue drivers this quarter?

🔍 Analysis Results:

1. **Product Category A** - $2.4M (40% of total)
   - 15% growth vs last quarter
   - Driven by enterprise segment

2. **Geographic Expansion** - $1.8M (30% of total)
   - New markets in APAC region
   - 3x growth vs last quarter

3. **Upsell to Existing Customers** - $1.2M (20% of total)
   - 25% conversion rate on upgrade offers
   - Average deal size: $50K

📈 Confidence: High (n=1,247 transactions)
⚠️  Caveats: Q4 includes holiday seasonality
💡 Recommendation: Double down on APAC expansion and enterprise upsells

Methodology

1. Decision First

• Identify the decision owner and question
• Clarify what would change based on results
• Set deadline before computing

2. Statistical Rigor

• Check sample size sufficiency
• Ensure fair comparison groups
• Account for multiple comparisons
• Quantify uncertainty (confidence intervals)
• Verify effect size is meaningful

3. Output Standards

• Lead with insight, not methodology
• Quantify uncertainty (ranges, not point estimates)
• State limitations clearly
• Recommend next steps

Security

⚠️ Data Transmission Warning

This skill reads the entire content of your data file and sends it to api.evolink.ai for analysis. Do not use this skill on files containing:

• API keys, tokens, or credentials
• Personally Identifiable Information (PII)
• Confidential business data
• Any sensitive information you don't want transmitted to an external service

The script implements security checks (directory constraints, symlink rejection, filename blacklist, size/MIME validation), but cannot guarantee that arbitrary data files are free of secrets.

Credentials & Network

Requires EVOLINK_API_KEY to call EvoLink API. Your data file content and analysis question are sent to api.evolink.ai for processing. EvoLink processes the data and returns analysis results. No data is stored after processing.

File Access

This skill reads the specified data file (CSV, Excel, JSON) from your local filesystem. Files must be within DATA_ANALYSIS_SAFE_DIR (default: $HOME/.openclaw/workspace).

Security validations:

• Path resolution via realpath -e (requires file to exist, resolves symlinks)
• Symlink inputs are explicitly rejected
• Directory constraint with trailing-slash comparison
• Filename blacklist: .env*, *.key, *.pem, *.p12, *.pfx, id_rsa*, authorized_keys, config.json, .bash_history, .ssh, shadow, passwd
• File size limit: 50MB maximum
• MIME validation: Only text/csv, text/plain, application/json, application/vnd.ms-excel, application/vnd.openxmlformats-officedocument.spreadsheetml.sheet accepted

Network Access

• EvoLink API (api.evolink.ai) - Sends data and receives analysis

All network calls use curl and can be audited in the script source.

Persistence & Privilege

This skill does not modify other skills or system settings. Does not request elevated or persistent permissions.

Links

• GitHub
• API Reference
• Community
• Support