http_get skill

Overview

This skill fetches a URL using curl and writes the response body to the out/ directory. It runs a simple shell command to retrieve the resource and saves it as a file you specify. Use it when you need a reproducible, scriptable way to download web content within an agent workflow.

How this skill works

The skill executes a bash script that invokes curl with the provided URL and target path. The HTTP response body is written to out/<filename> while preserving basic curl behavior (redirects, status codes, etc.). The script returns the downloaded file path so calling code can read or process the saved content.

When to use it

• Quickly download a webpage or raw resource for processing
• Capture HTML or text output for offline analysis or parsing
• Integrate into pipelines that need a local copy of a URL resource
• Automate retrieval of assets before running downstream tools
• Test agent integrations that require fetching external content

Best practices

• Provide a clear filename under out/ to avoid collisions and to make outputs predictable
• Validate the URL before calling the script to avoid unnecessary network calls
• Check curl exit codes and HTTP status in downstream logic to handle failures gracefully
• Sanitize or limit file sizes if fetching untrusted sources to avoid disk exhaustion
• Use network and timeout controls (via curl options) for robust automation

Example use cases

• Download a webpage and run an HTML parser on out/example.html
• Fetch a JSON endpoint and pass the saved file to a validator or transformer
• Automate pre-processing in an agent pipeline that needs external content
• Retrieve remote training examples or prompts for offline model evaluation
• Quickly capture a resource for debugging an integration

FAQ