Document Desk Agent is a WebMCP-powered document workflow assistant for PDFs and scanned documents.
The goal is to reduce the repetitive manual work people face when handling invoices, letters, contracts, certificates, reports, and other documents.
The app analyzes an uploaded document and proposes structured actions such as:
- detecting the document type
- generating a useful filename
- suggesting a filing location
- extracting tasks and deadlines
- creating a short summary
- highlighting important information
The key design principle is human control. The assistant can prepare actions, but the user reviews and approves them before anything is applied.
WebMCP is used to expose these document actions directly to the agent as structured tools instead of forcing the agent to interact with the interface by clicking buttons visually.
This creates a workflow where the agent understands the available document operations, prepares useful changes, and collaborates with the user through explicit approval steps.
Built With
- css
- github
- html
- javascript
- openai
- pdf.js
- webmcp
Log in or sign up for Devpost to join the conversation.