Anthropic Claude Batch Prompt Processor and Poller - n8n Workflow

Users needing to process hundreds or thousands of AI prompts without hitting rate limits (Anthropic Batch API users).
Automation Specialists looking for robust n8n templates featuring API polling logic.
Developers building scalable AI applications based on Anthropic Claude.
Anyone seeking a powerful n8n workflow solution for complex AI data handling.

Processing a large volume of AI requests synchronously can lead to timeouts and rate limit errors. This solution leverages the Anthropic Batch API, which is ideal for massive processing jobs. This specific n8n workflow acts as a reliable automation wrapper: it submits your array of prompts, initiates a polling sequence, and waits intelligently for the job to complete. This specific n8n template includes examples demonstrating how to structure requests, including integrating context from a Chat History n8n node. This approach makes this an essential piece of core logic flow control for any serious n8n user dealing with scalable AI operations. This entire n8n workflow ensures efficient resource utilization and reliable result retrieval.

The core of this n8n workflow is designed to be called by an external automation or trigger.

1. Trigger and Submission: The process starts with the 'When Executed by Another Workflow' n8n trigger, expecting the API version and the array of AI prompts (requests). This data is immediately passed to the 'Submit batch' n8n node, which uses an HTTP Request to initiate the batch job with the Anthropic API.


2. Polling Loop: The 'If ended processing' n8n node checks the status of the batch job. If the processingstatus is 'ended', the flow proceeds to retrieval. If not, the flow enters a wait-and-check loop. The 'Batch Status Poll Interval' n8n node pauses the n8n workflow for 10 seconds before the 'Check batch status' n8n node sends another request to monitor the job's progress. This loop continues until the batch is complete.


3. Result Retrieval and Parsing: Once processing is 'ended', the 'Get results' n8n node retrieves the data using the provided resultsurl. The results are typically in JSONL format. The 'Parse response' Code n8n node then executes a custom script to split the newline-separated JSONL data into individual, structured JSON items.


4. Final Output: The 'Split Out Parsed Results' n8n node ensures that each prompt result from the batch is returned as a separate item, ready for further processing in your larger n8n workflow.

To use this robust n8n workflow, follow these steps:

1. Import: Copy the provided JSON code and import it into your n8n instance using the 'New' menu and selecting 'Import from JSON'.


2. Credentials: This n8n workflow requires Anthropic API credentials. You must set up an 'Anthropic Api' credential set in n8n. Ensure the credential is selected in the 'Submit batch' and 'Check batch status' n8n nodes.


3. Chat Memory Setup (Optional Example): If you intend to use the embedded example demonstrating Chat History, you must configure a 'Simple Memory Store' n8n node (Memory Buffer Window) and ensure the subsequent LangChain nodes ('Fill Chat Memory with example data', 'Load Chat Memory Data') are correctly linked to it.


4. Execution: The primary flow ('When Executed by Another Workflow' n8n trigger) is designed to be called by an 'Execute Workflow' n8n node from another parent workflow, providing the required requests array and anthropic-version header.

This high-performing n8n workflow leverages several specialized n8n node types:

When Executed by Another Workflow (n8n trigger): This is the entry point, defining the expected input structure (anthropic-version, requests array) required to run this segment of the n8n workflow.
Submit batch (HTTP Request n8n node): Posts the aggregated batch prompts to https://api.anthropic.com/v1/messages/batches. Key configuration includes setting the anthropic-version header dynamically and sending the batch requests array as the JSON body.
If ended processing (If n8n node): Provides core logic flow control. It uses a dynamic expression (={{ $json.processing_status }}) to check if the status equals 'ended'. This branching is crucial for the polling loop.
Batch Status Poll Interval (Wait n8n node): A critical n8n node for API polling, pausing execution for 10 seconds to avoid overloading the API while waiting for job completion.
Check batch status (HTTP Request n8n node): Sends follow-up requests using the job ID to monitor the batch progress until it completes.
Get results (HTTP Request n8n node): Retrieves the final result file URL supplied by the Anthropic API once the batch job is marked as complete.
Parse response (Code n8n node): Contains a JavaScript script specifically designed to handle the Anthropic JSONL file format, splitting the response data by newlines and parsing each line into a standard JSON object. This custom code is essential for processing the raw output.
Split Out Parsed Results (Split Out n8n node): Takes the parsed array of results and converts each item into a separate execution item, allowing subsequent n8n nodes to process results individually.

• Fill Chat Memory with example data / Load Chat Memory Data (LangChain Memory Manager n8n node): Used in the example flow to demonstrate how to pull conversational history from a Simple Memory Store n8n node and construct a context-aware batch request object for Claude.