CrunchAPI - API Documentation

One endpoint. Send CSV data and plain-English instructions. Get structured analysis back.

POST /v1/analyze

Upload a CSV file with natural language instructions. CrunchAPI generates analysis code, executes it in a sandbox, self-heals on errors (up to 3 retries), and returns structured results with a human-readable summary.

Request (multipart/form-data)

Field	Type	Description
csv required	file	CSV file to analyze (max 10MB)
instructions required	string	Natural language description of the analysis to perform

Example Request (cURL)

curl -X POST https://crunchapi-ndpj.polsia.app/v1/analyze \
  -F csv=@sales_q4.csv \
  -F instructions="Find the top 3 products by revenue growth rate. Show month-over-month trends."

Example Request (Python)

import requests

response = requests.post(
    "https://crunchapi-ndpj.polsia.app/v1/analyze",
    files={"csv": open("sales_q4.csv", "rb")},
    data={"instructions": "Find top 3 products by revenue"}
)
print(response.json())

Example Request (JavaScript)

const form = new FormData();
form.append('csv', fileInput.files[0]);
form.append('instructions', 'Find top 3 products by revenue');

const res = await fetch('/v1/analyze', {
  method: 'POST',
  body: form
});
const data = await res.json();

Response

Field	Type	Description
summary	string	Natural language summary of the analysis findings
results	object\|array	Structured analysis data
code_executed	string	The analysis code that was generated and executed
retries	number	Number of self-healing retries (0 = first attempt succeeded)
metadata	object	Request metadata (rows_analyzed, columns, duration_seconds)

Example Response

{
  "summary": "Widget Pro led Q4 with 34% MoM revenue growth. CloudSync and DataPack followed at 28% and 21%. Anomaly: Widget Pro spiked 180% in November.",
  "results": {
    "top_products": [
      { "product": "Widget Pro", "growth_rate": 0.34, "total_revenue": 82500 },
      { "product": "CloudSync", "growth_rate": 0.28, "total_revenue": 59800 }
    ]
  },
  "code_executed": "const grouped = data.reduce(...);\nresult = { top_products: ... };",
  "retries": 0,
  "metadata": {
    "rows_analyzed": 1847,
    "columns": ["product", "month", "revenue"],
    "duration_seconds": 2.3
  }
}

Status	Meaning	When
200	Success	Analysis completed
400	Bad Request	Missing CSV, bad format, or missing instructions
500	Analysis Failed	Code generation/execution failed after all retries

API Reference

Try It Now