> ## Documentation Index
> Fetch the complete documentation index at: https://docs.declaw.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Download Results

> Generate multiple output files inside a Declaw sandbox and iterate over them to read them back to the host.

## What You'll Learn

* Creating directories with `sbx.commands.run("mkdir -p ...")`
* Writing and executing a file-generating script
* Listing output files with `sbx.files.list()`
* Reading multiple files with `sbx.files.read()` in a loop
* Iterating over `EntryInfo` objects (`name`, `type`, `size`, `path`)

## Prerequisites

<Snippet file="snippets/env-setup.mdx" />

<Note>This example is available in Python. TypeScript version coming soon.</Note>

## Code Walkthrough

**Create an output directory** inside the sandbox:

```python theme={null}
from declaw import Sandbox

sbx = Sandbox.create(template="python", timeout=300)
try:
    sbx.commands.run("mkdir -p /tmp/output")
    print("Created /tmp/output")
```

**Write and run a generator script** that produces a JSON report and a CSV:

```python theme={null}
    generator_script = '''
import json
import datetime

report = {
    "generated_at": datetime.datetime.now().isoformat(),
    "status": "complete",
    "metrics": {
        "accuracy": 0.95,
        "precision": 0.93,
        "recall": 0.97,
        "f1_score": 0.95
    },
    "summary": "Model evaluation complete. All metrics above threshold."
}

with open("/tmp/output/report.json", "w") as f:
    json.dump(report, f, indent=2)

with open("/tmp/output/predictions.csv", "w") as f:
    f.write("id,predicted,actual,correct\\n")
    for i in range(10):
        pred = i % 3
        actual = i % 3 if i != 7 else (i % 3 + 1) % 3
        f.write(f"{i},{pred},{actual},{pred == actual}\\n")

print("Files generated successfully!")
'''
    sbx.files.write("/tmp/generate.py", generator_script)
    result = sbx.commands.run("python3 /tmp/generate.py")
    print(f"stdout: {result.stdout}")
```

**List the output directory** — each `entry` has `name`, `type`, `size`, and `path`:

```python theme={null}
    entries = sbx.files.list("/tmp/output")
    for entry in entries:
        print(f"  {entry.name} ({entry.type}, {entry.size} bytes)")
```

**Read each file back** using the `path` attribute from the directory listing:

```python theme={null}
    for entry in entries:
        print(f"\n--- Reading {entry.name} ---")
        content = sbx.files.read(entry.path)
        print(content)
finally:
    sbx.kill()
```

## Expected Output

```
==================================================
Declaw Download Results Example
==================================================

--- Creating Sandbox ---
Sandbox created: sbx_abc123

--- Creating Output Directory ---
Created /tmp/output

--- Writing Generator Script ---
Wrote /tmp/generate.py

--- Running Generator Script ---
stdout: Files generated successfully!

--- Listing Output Files ---
  report.json (file, 245 bytes)
  predictions.csv (file, 178 bytes)

--- Reading report.json ---
{
  "generated_at": "2026-04-02T12:00:00.000000",
  "status": "complete",
  "metrics": {
    "accuracy": 0.95,
    "precision": 0.93,
    "recall": 0.97,
    "f1_score": 0.95
  },
  "summary": "Model evaluation complete. All metrics above threshold."
}

--- Reading predictions.csv ---
id,predicted,actual,correct
0,0,0,True
1,1,1,True
...

--- Cleaning Up ---
Sandbox killed.

==================================================
Done!
==================================================
```