# Offloading to Files and Memory — Agent Loops, Stop Conditions and Token Budgets

Source: https://www.geekswithgeeks.com/en/agent-loops/ctx-offload

> Store bulky data outside the context and read it back by reference.

## Keep a pointer, not the data

Instead of keeping a 20,000-token document in the conversation, save it to a file and keep only its path and a two-line description in context. The agent reads the file, or a part of it, only when needed. The same idea works for a notes file where the agent records findings so they survive trimming and restarts.

## A tool that returns a reference

The big content goes to disk. The model sees a short handle and can call `read_chunk` for a specific slice.

```python
def fetch_report(url: str) -> str:
    path = save_to_workspace(download(url))
    return f"Saved to {path} ({size_kb(path)} KB). Use read_chunk(path, start, length)."
```

**Quiz:** What stays in context when you offload a large document to a file?

- [ ] The whole document
- [x] Only a short reference such as the path and description
- [ ] Nothing at all
- [ ] A copy for every turn

*Answer:* Only a short reference such as the path and description. A small handle keeps context lean while the data stays available on demand.