Extend Block

Parse documents into structured markdown or spatial content and retrieve processor run results through the Extend AI API. Authenticate with an Extend API key to unlock document parsing and run-status retrieval in any workflow.

Overview

When to Use

• Convert a PDF, Word file, or any document at a public URL into structured markdown for downstream text processing.
• Extract spatial/layout-aware content from documents where positional information matters.
• Poll an Extend processor run by ID to check its status or fetch its completed output.
• Build document-ingestion pipelines that feed parsed content into an agent or knowledge base.
• Choose a specific parsing engine (parse_performance or parse_light) to balance speed versus quality.
• Apply page-, document-, or section-level chunking strategies to align output size with your downstream needs.

Configuration

Operation

Selects which Extend API action to perform. The fields shown in the block change based on this selection.

Default: extend_parse

File URL (shown when operation = extend_parse)

• Type: short-input
• Placeholder: https://example.com/document.pdf
• The publicly accessible URL of the document to parse. Must be reachable by Extend's servers.

Output Format (shown when operation = extend_parse)

• Type: short-input
• Placeholder: markdown
• Target output format for the parsed document. Accepted values: markdown (default) or spatial.

Chunking Strategy (shown when operation = extend_parse)

• Type: short-input
• Placeholder: page
• Controls how the parsed content is split. Accepted values: page, document, or section.

Engine (shown when operation = extend_parse)

• Type: short-input
• Placeholder: parse_performance
• Parsing engine to use. Accepted values: parse_performance (higher quality) or parse_light (faster).

Run ID (shown when operation = extend_get_run)

• Type: short-input
• Placeholder: run_...
• The ID of the Extend processor or parse run to retrieve. Returned by a prior extend_parse call as {{extend_1.metadata.id}}.

Extend API Key (always shown, required)

• Type: short-input (password)
• Required: yes
• Your Extend API key used to authenticate all requests. Store it as a workflow secret and reference it as {{EXTEND_API_KEY}}.

Inputs & Outputs

Inputs

• operation (string) — Operation to perform (extend_parse or extend_get_run).
• apiKey (string) — Extend API key.
• fileUrl (string) — URL of the document to parse (used by extend_parse).
• outputFormat (string) — Target output format (used by extend_parse).
• chunking (string) — Chunking strategy (used by extend_parse).
• engine (string) — Parsing engine (used by extend_parse).
• runId (string) — Processor run ID to retrieve (used by extend_get_run).

Outputs

• data (json) — Result object from Extend. For extend_parse this is the parsed document object; for extend_get_run this is the processor run object.
• metadata (json) — Response metadata containing id (string — run/parse ID) and status (string — processing status).

Tools

Extend Parse Document (extend_parse) — Submits a POST request to https://api.extend.ai/parse with the document URL and optional configuration (output format, chunking strategy, engine). Returns the parsed document object and run identifiers.

Extend Get Run (extend_get_run) — Issues a GET request to https://api.extend.ai/processor_runs/{runId} to retrieve the status and result of an existing Extend processor or parse run by its ID.

YAML Example

# Example 1: Parse a PDF document into markdown
extend_1:
  type: extend
  name: "Parse Document"
  inputs:
    operation: extend_parse
    apiKey: "{{EXTEND_API_KEY}}"
    fileUrl: "{{fetch_doc_block.url}}"
    outputFormat: markdown
    chunking: page
    engine: parse_performance
  connections:
    outgoing:
      - target: agent_1

# Example 2: Retrieve an Extend run by ID
extend_2:
  type: extend
  name: "Get Run Result"
  inputs:
    operation: extend_get_run
    apiKey: "{{EXTEND_API_KEY}}"
    runId: "{{extend_1.metadata.id}}"
  connections:
    outgoing:
      - target: condition_1