Documentation

Agent Files and Storage

Every agent in O-mega has its own file storage—a dedicated space where documents, downloads, generated content, and extracted data accumulate over time. This isn't just an archive; it's a working resource that your agent understands and can intellige...

Every agent in O-mega has its own file storage—a dedicated space where documents, downloads, generated content, and extracted data accumulate over time. This isn't just an archive; it's a working resource that your agent understands and can intelligently interact with.

When your agent downloads a PDF, generates a report, or extracts data from a website, the results land in this storage. But unlike a simple file folder, your agent maintains awareness of what's there. You can ask about files in natural language, request specific items by description, and use past outputs as inputs for new work.

File Sources

Files arrive in your agent's storage from various activities. Understanding where files come from helps you know what to expect in your agent's collection.

Files come from:

  • Browser downloads - PDFs, images, spreadsheets, and other files your agent downloads during web sessions
  • Generated content - Websites, documents, and data outputs created in computer sessions
  • Uploaded files - Files you share with your agent through the conversation
  • Extracted data - Tables, lists, and structured information gathered from websites

Over time, an active agent accumulates a rich library of materials related to its work. This library becomes a resource for future tasks.

Intelligent File Management

What sets O-mega's file system apart is that your agent truly understands its files. You don't need to remember exact filenames or browse through folders—you can describe what you need, and your agent will find it.

Intelligent capabilities include:

  • Search by content - Ask your agent to find files about a specific topic, and it will locate relevant items based on what they contain, not just their names
  • Filter and organize - Request files by type, date range, or relevance to a particular project
  • Use in tasks - Reference previous files when giving new instructions, and your agent will incorporate them

For example, you might say "Use the competitor analysis from last week to update the strategy document" and your agent knows exactly which file you mean. Or "Find all the spreadsheets I downloaded last month" and your agent filters appropriately.

This intelligence means you can work with files conversationally rather than through file managers and folder navigation.

Files Section

For direct access to files, every agent has a Files section in its interface. This is where you can browse, preview, and manage everything your agent has collected.

In the Files section you can:

  • View a complete list of all files your agent has access to
  • Preview files directly in the interface without downloading
  • Download files to your local device
  • See when each file was created or modified
  • Delete files you no longer need

The Files section is organized chronologically by default, with the most recent files appearing first. This makes it easy to find work from recent sessions.

Using Files in Conversations

Files are a first-class part of your agent conversations. You can both send files to your agent and work with files your agent has created.

When chatting with your agent:

  • Attach files to your messages for your agent to analyze, transform, or incorporate
  • Ask about files your agent has created ("What's in that spreadsheet you made yesterday?")
  • Request modifications to existing files ("Update the Q3 report with the new numbers")
  • Reference past work naturally ("Like we did in the marketing deck")

Your agent maintains context about its file collection. When you reference "the report from last week" or "that spreadsheet with customer data," your agent can usually identify what you mean without requiring exact filenames.

This conversational file interaction eliminates the friction of traditional file management. You work with your agent as you would with a knowledgeable colleague who remembers where everything is.

Related: Understanding Deliverables | Computer Sessions Overview