TollNode

CSV checker and converter API for AI agents

TollNode gives Claude, Cursor or any AI agent a pre-import check for CSV, JSON and Excel files (bad rows, broken emails, duplicates, a score) and conversion between CSV, JSON and .xlsx, from $0.003 per file.

Ask in plain words, like 'check this export before I import it into the CRM', and get the problems with row numbers.

No subscription · Sign-up optional · From $0.003 per call · Charged only when it works · No account needed

Claude Code

Check this contacts CSV before I import it. Emails must be unique

TollNode · preflight$0.003 · 4 rows · 1 s

CheckResult
Verdictfix first (score 50 of 100)
invalid_email1 row: row 3
duplicate_emails1 row: row 4
leading_zeros in zipinfo: import as text

Real result.

Works in Claude Code, Cursor, Codex, VS Code and any MCP agent.

What your agent gets from a data file

Every result is structured JSON your agent can sort, filter, summarize or turn into a spreadsheet.

What people ask their AI to do with it

Use it as a CSV checker MCP server in Claude, Cursor or Codex

Connect once. Then just ask.

Paste this into Claude Code, Cursor or Codex. It connects TollNode and gives your agent a small wallet. Add a few dollars of USDC and you're done.

terminal
npx agentcash add https://tollnode.com

Copy for your AI agent

Paste this into your agent so it knows what this node does and how to call it.

prompt
You can use TollNode's Data file check and convert node. Call POST https://tollnode.com/api/v1/nodes/data-file with an "action" (preflight, convert) and its fields: to (optional), url (optional), text (optional), sheet (optional), format (optional), delimiter (optional), unique_column (optional), required_columns (optional). Each action has its own price and is charged only when it returns a result. Over MCP, call the "data-file" tool; use list_models to see prices. Return results as tables when I ask for lists.

2 actions, one node

Send action plus the fields that action needs, to the same address. Each action has its own price, charged only when it returns a result. Leave out action and you get preflight.

  • preflight$0.003

    Checks a file the way an importer would and lists everything that will break. Verdict, score, issues with row numbers, and each column's type, fill rate, unique count and examples.

    needs: url or text · optional: required_columns, unique_column, format, sheet

  • convert$0.003

    Converts between CSV, JSON and Excel (.xlsx). CSV as content, JSON as data (numbers typed, ZIP-style codes kept as text), or .xlsx as a dataUrl with a filename.

    needs: url or text, to · optional: delimiter, format, sheet

request body: preflight
{"action":"preflight","url":"https://example.com/customers.csv"}
request body: convert
{"action":"convert","url":"https://example.com/customers.csv","to":"json"}

POST either body to https://tollnode.com/api/v1/nodes/data-file. From an AI assistant you don't write this yourself: ask in plain words and it picks the action.

Charged only when a call returns a result. Failed or empty calls cost $0.

Check and convert CSV and Excel files without a script

FAQ

Is there an official tool for this?
No. This is TollNode's own checker and converter: it reads the file the way an importer would and lists what will break.
Can I use it as a Data file check and convert MCP server in Claude?
Yes. Add https://tollnode.com/mcp in Claude Code, Claude Desktop or Cursor, and the "data-file" tool appears with its fields and price.
Which files work?
CSV, TSV, JSON or .xlsx from a public link (up to 10 MB; Excel up to 5 MB), or paste the text. Public Google Sheets share links work. Old .xls files, PDFs and Word files are not supported.
What does preflight report?
A verdict, a 0 to 100 score, every issue with severity, column and Excel row numbers, and a profile of each column.
What does convert return?
CSV as text, JSON as records with numbers typed, or an .xlsx file as a data URL with a filename.
Do I need an account with another service?
No. Your agent calls TollNode, and you never log in to another service or share cookies.
What if nothing comes back?
You're not charged. TollNode only collects payment when a usable result is returned, and your agent can try a different input.

Ready in a minute

Connect your AI
For developers: endpoint, inputs and an example response

POST https://tollnode.com/api/v1/nodes/data-file

Input

FieldTypeNotes
toenumconvert: the format you want (csv, json, xlsx)
urlstringA public link to a CSV, TSV, JSON or .xlsx file (up to 10 MB; Excel up to 5 MB). Google Sheets share links work when the sheet is public
textstringOr paste the CSV or JSON itself
sheetstringExcel sheet name. Default: the first sheet with data
action*enumpreflight checks the file; convert turns it into another format (preflight, convert)
formatenumOnly if detection guesses wrong (csv, tsv, json, xlsx)
delimiterenumconvert to csv: the separator. Default comma (comma, semicolon, tab)
unique_columnstringpreflight: a column whose values must not repeat, e.g. email or sku
required_columnsarraypreflight: columns that must exist and be filled in every row

Example response

{
  "file": {
    "name": "customers.csv",
    "bytes": 88213,
    "format": "csv",
    "encoding": "UTF-8",
    "delimiter": "comma"
  },
  "note": "Row numbers count the header as row 1, as in Excel. Each issue lists up to 20 rows; count has the total.",
  "rows": 1204,
  "score": 55,
  "issues": [
    {
      "rows": [
        14,
        88,
        301
      ],
      "check": "invalid_email",
      "count": 7,
      "column": "email",
      "message": "Values in 'email' that are not valid email addresses.",
      "severity": "error"
    },
    {
      "rows": [
        57,
        912
      ],
      "check": "wrong_field_count",
      "count": 2,
      "message": "2 rows have 7 fields; the header has 6. Usually an unquoted comma or a missing value.",
      "severity": "error"
    },
    {
      "rows": [
        40,
        77
      ],
      "check": "duplicate_emails",
      "count": 12,
      "column": "email",
      "message": "The same email appears more than once in 'email' (ignoring case).",
      "severity": "warning"
    },
    {
      "rows": [
        2,
        3,
        4
      ],
      "check": "date_format",
      "count": 1204,
      "column": "signup",
      "message": "Dates in 'signup' are not in YYYY-MM-DD form. Importers can swap day and month (03/04 is March or April).",
      "severity": "warning"
    },
    {
      "rows": [],
      "check": "leading_zeros",
      "count": 96,
      "column": "zip",
      "message": "'zip' has codes with leading zeros (like ZIP codes). Import it as text, or the zeros are lost.",
      "severity": "info"
    }
  ],
  "columns": [
    {
      "name": "email",
      "type": "email",
      "empty": 6,
      "filled": 1198,
      "unique": 1186,
      "examples": [
        "ana@example.com",
        "ben@shop.io",
        "cy@mail.com"
      ]
    },
    {
      "name": "zip",
      "type": "text",
      "empty": 0,
      "filled": 1204,
      "unique": 611,
      "examples": [
        "02134",
        "10001",
        "94105"
      ]
    }
  ],
  "summary": "1,204 rows, 6 columns. 2 errors, 3 warnings.",
  "verdict": "fix first"
}

Also described in the OpenAPI file and llms.txt. Setup for every app: Agents & MCP.