Read a URL File as Text in Jinba Flow

Use INPUT_FILE_AS_TEXT when a flow needs the contents of a file available at a direct URL. The value input accepts one URL or an array of URLs. The result is the file content as text: one string for a single URL, or an array for multiple URLs. A one-element array returns one string.

Fetch the file content

- id: read_file
  tool: INPUT_FILE_AS_TEXT
  input:
    - name: value
      value: "https://example.com/config.json"
    - name: description
      value: "Configuration to process"

Replace the URL with a reachable direct file URL. On success, inspect result for the text you expect. The file picker lists plain text, CSV, HTML, XML, JSON, and YAML types; this tool returns text rather than parsing those formats. description is optional. use_as_input defaults to true, and optional defaults to false.

Use the returned text

A later step can consume the returned text directly; parse JSON or CSV explicitly when you need fields or rows:

- id: summarize_file
  tool: OPENAI_INVOKE
  needs: [read_file]
  input:
    - name: prompt
      value: |
        Summarize the content of this file. If it is CSV, identify its columns.
        {{steps.read_file.result}}

For a CSV workflow that needs parsed rows, use INPUT_CSV_FILE instead; see Summarize a CSV file. Avoid putting API keys or other secrets from a configuration file into an AI prompt.

Check a JSON configuration file

Parse the text in Python and check for required keys without printing their values:

- id: read_config
  tool: INPUT_FILE_AS_TEXT
  input:
    - name: value
      value: "https://example.com/config.json"

- id: check_config
  tool: PYTHON_SANDBOX_RUN
  needs: [read_config]
  input:
    - name: data_type
      value: OBJECT
    - name: code
      value: |
        import json

        text = r'''{{steps.read_config.result}}'''.strip()
        required = ["database_url", "api_key"]

        try:
            config = json.loads(text) if text else None
            error = None if isinstance(config, dict) else "Config must be a JSON object"
        except json.JSONDecodeError as e:
            config, error = None, f"Invalid JSON: {e}"

        missing = [key for key in required if not config.get(key)] if error is None else []

        print(json.dumps({"valid": error is None and not missing, "error": error, "missing": missing}))

steps.check_config.result.data.valid is true only when the file is a JSON object with every required key set. Otherwise result.data.error explains a parse problem or result.data.missing lists keys that are absent or empty. Edit required to match your file. To reject bad JSON sent to a published workflow instead, use Input JSON with Validation.

Check file input failures

If no URL is supplied, the step fails with “File URL is required.” A failed fetch reports its HTTP status. Check the URL, access permissions, and response status. A file larger than 100 MiB, a download that takes longer than 5 minutes, or more than 50 URLs in one step also fails the step (see Download limits). If optional is true and no URL is supplied, result is null.

​Fetch the file content

​Use the returned text

​Check a JSON configuration file

​Check file input failures

Fetch the file content

Use the returned text

Check a JSON configuration file

Check file input failures