upload_data

Upload a CSV or Excel file as a named data source. IMPORTANT: file_content must be actual base64-encoded file bytes — not a filename, path, or prose description. Obtain it from a file-read tool, a client upload, or a local script (e.g. base64.b64encode(open(‘file.csv’, ‘rb’).read()).decode()). Typical flow: (1) read the file bytes via a file-read tool or upload mechanism, (2) base64-encode them, (3) pass the encoded string here. Passing anything other than real base64-encoded bytes produces a clear error identifying the sourcing problem. The file is converted to Parquet and stored in S3 under {tenant_id}/{project_name}/uploads/{source_name}.parquet. Maximum file size: 100 MB. Schema is inferred automatically with best-effort type coercion. Accepted formats: .csv, .xlsx, .xls. Returns: source_name (str) — data source identifier; project_name (str) — project name; filename (str) — original uploaded filename; s3_key (str) — Parquet file location in S3; row_count (int) — number of rows ingested; columns (int) — number of columns; schema (dict[str, str]) — column-name to data-type mapping; parquet_size_bytes (int) — size of the stored Parquet file.

Parameters

Name	Type	Required	Default	Description
project_name	str	Yes	—
source_name	str	Yes	—
file_content	Annotated[str, Field(max_length=MAX_FILE_CONTENT_CHARS, description='Base64-encoded binary content of the file. Must be obtained from an external source — a file-read tool result, a client upload, a local script, or a prior tool output that returns base64 content. Do NOT pass a filename, path, or text description.')]	Yes	—
filename	str	Yes	—

Returns

Returns a structured object. Top-level fields:

• source_name (string)
• project_name (string)
• filename (string)
• s3_key (string)
• row_count (integer)
• columns (integer)
• schema (object)
• parquet_size_bytes (integer)

See the response example below for the full payload shape.

Example

Tool Call

{
  "name": "upload_data",
  "arguments": {
    "project_name": "forecasts",
    "source_name": "actuals",
    "file_content": "PEJBU0U2NCBPRiAuY3N2IEZJTEUgQ09OVEVOVD4=",
    "filename": "actuals_2026q1.csv"
  }
}

Response

{
  "source_name": "actuals",
  "project_name": "forecasts",
  "filename": "actuals_2026q1.csv",
  "s3_key": "tenants/acme/forecasts/revenue/data.parquet",
  "row_count": 24,
  "columns": 4,
  "schema": {
    "month": "DATE",
    "revenue": "DOUBLE",
    "cogs": "DOUBLE",
    "headcount": "BIGINT"
  },
  "parquet_size_bytes": 2048
}

Notes

• Module: Data
• Requires authentication via API token or Auth0 JWT.