Adobe PDF Services jobs run from a reusable workflow

Adobe PDF Services jobs run from a reusable workflow

An n8n workflow that other workflows can call to run an Adobe PDF Services operation on a PDF. You pass it an operation name, the settings of that operation and the file. It gets a token, uploads the PDF, starts the job, waits for it and returns Adobe’s answer. It holds no credential.

What is inside

• workflow.json: the workflow as n8n exports it, 14 nodes beside 6 sticky notes. It holds no credential.
• README.md: these notes.

What we changed from the original copy

• The test node Load a test pdf file pointed at a PDF in the original author’s own Dropbox folder. Mindket replaced the path with /<YOUR_DROPBOX_FOLDER>/<YOUR_FILE>.pdf.
• The token node had a typo in its name. Mindket renamed it Get token and updated the two nodes that read its answer.
• The sticky notes are reworded. The note for extractpdf named splitpdf as its endpoint, and the split example had one closing brace too many. The new note holds one correct example, for extractpdf, and the credential notes use placeholders where the original showed masked stars.
• No credential is inside.

Before you start

• An n8n instance, on n8n Cloud or your own server.
• An Adobe PDF Services project, with its client ID and client secret.
• For the test part, a Dropbox account and a PDF in it.

Set it up

1. In n8n, create a workflow, open its menu and choose Import from File. Pick workflow.json.
2. Create a Custom Auth credential with the JSON in the note next to Get token. Put your client ID and client secret in it and select it in Get token.
3. Create a Header Auth credential with the name X-API-Key and your client ID as its value. Select it in Create Asset, Process Query and Try to download the result.
4. To test by hand, add a Dropbox credential, select it in Load a test pdf file and put the path of a PDF in its Path field.
5. Choose Test workflow and read the answer in Forward response to origin workflow.

How a run goes

1. Execute Workflow Trigger takes a call from another workflow. In a test, When clicking ‘Test workflow’ starts it, Adobe API Query sets endpoint to extractpdf with a request for text and tables, and Load a test pdf file fetches the PDF.
2. Query + File joins the query and the file by position. Get token posts to Adobe’s token address and gets an access token.
3. Create Asset asks for a place to put the file and gets uploadUri and assetID. Query + File + Asset information joins them with the query and the file.
4. Upload PDF File (asset) sends the PDF to uploadUri with a PUT.
5. Process Query posts assetID and json_payload to the operation address. Adobe answers with the job’s address in the location header.
6. Wait 5 second pauses, and Try to download the result reads the job’s status. Switch sends in progress back to the wait. It sends failed and every other status on.
7. Forward response to origin workflow hands the last answer to the caller. A finished job holds a link to the result.

Change the look

• The operation: the endpoint field of Adobe API Query, or the input from the calling workflow.
• The settings of the operation: the json_payload field of Adobe API Query.
• The test PDF: the path in Load a test pdf file.
• How often it checks the job: the seconds in Wait 5 second.

Good to know

• Nothing starts it by itself. Only the test button or a calling workflow runs it.
• Nothing limits the checking rounds. A job that never finishes keeps the run open, so set a timeout in the workflow settings.
• A finished job holds a link to the result file. This workflow does not fetch that file, so do it in your own workflow.
• Adobe’s documentation, read on 12 November 2026, says to keep the client secret in a safe server environment. Keep it in the n8n credential only.
• Adobe’s own limits and prices apply to each job.
• Mindket did not run this workflow. It read Adobe’s getting-started page for the steps and n8n’s source for the nodes.
• The JSON holds no credential.