- READY
- ACTIVE
- HUMAN
- LOGGED
Short answer
How can I start document processing as soon as files are uploaded?
You tell your agent to run a workflow execution once it detects new files in your remote storage. This kicks off the Unstructured ingestion process so your data is ready for analysis right away. You'll see how the agent handles the handoff below.
The results.
Where your data lands.
The ingestion process moves files from storage into a format your agent can actually use.
READY
Cleaned data
The agent receives structured text and metadata from the files. You can immediately ask questions about the content.
ACTIVE
Running jobs
The agent monitors the status of the ingestion. You get updates on whether the files processed successfully.
HUMAN
Manual review
If a file format is unsupported or corrupted, the agent flags it. You'll need to step in to fix the source file.
LOGGED
Execution history
Every triggered job is tracked. You can verify which files were ingested and when the process finished.
The workflow
What your AI does when the files arrive.
The agent acts as the bridge between your storage and the Unstructured processing engine.
Detect new files
Your agent checks your remote source for any new uploads that haven't been processed yet.
trigger_workflow_executionStart ingestion
The agent sends the file paths to Unstructured to begin the extraction process.
trigger_workflow_executionMonitor progress
The agent tracks the job to ensure the files are being parsed correctly.
trigger_workflow_executionDeliver content
Once finished, the agent makes the extracted text available for your prompts.
trigger_workflow_execution
Try it
Copy these to start.
Use these prompts to kick off the process in your chat interface.
Starting points
Swap out the bracketed text for your specific folder names or file types.
Unstructured Connector
You're all set. Choose your MCP client and follow the setup instructions.
https://edge.vinkius.com/vk_preview_wLKl0zBun6GGEWOAXV2AtIEfq4BCwiQMTDSCBvt6/mcpClaude Desktop
Follow the steps below to connect in seconds.
- 1In Claude Desktop, open Settings → Connectors.
- 2Click “Add custom connector” and paste the connector link above as the remote MCP server URL.
- 3Click Add and start a new chat — Unstructured capabilities are ready to use.
{
"mcpServers": {
"unstructured-mcp": {
"url": "https://edge.vinkius.com/vk_preview_wLKl0zBun6GGEWOAXV2AtIEfq4BCwiQMTDSCBvt6/mcp"
}
}
}Trigger a new ingestion job for the files in my S3 bucket.
Start processing the PDFs I just uploaded to my Google Drive folder.
Check if the latest documents in my Dropbox are ready for ingestion.
Run the ingestion workflow for the new files in my Azure blob storage.
Claude
ChatGPT
Cursor
VS Code
Windsurf
Claude Code
JetBrains
Cline
Start here
Connect Unstructured once, then ask.
Log in once to link your account. Your credentials stay encrypted, and you can find the full tool list on the Connector page.
Connect Unstructured to your AIFAQ
How the ingestion works
- 01
Can the AI delete files from my S3 bucket?
No. The agent can only trigger the ingestion workflow; it cannot delete or modify your original files.
- 02
Does the AI automatically know when a file is uploaded?
The AI triggers the process when you ask it to or when you set up a trigger. It doesn't watch your folders in the background without an instruction.
- 03
What happens if the file is a weird format?
If Unstructured can't parse the file, the workflow will fail. The agent will tell you it couldn't process the document.
- 04
Can I use this to edit the documents?
No. This tool is for reading and extracting data, not for writing or editing files in your storage.
- 05
Can the agent process multiple files at once?
Yes, you can trigger the workflow for a batch of files in a specific directory.
More questions about Unstructured? The Connector page answers them. See everything the Unstructured Connector can do
Connect Unstructured to Claude, Cursor, ChatGPT & more
