The Agent Skills Directory

[PROMPT_INJECTION]: The skill fetches and processes content from external blog URLs and RSS feeds, which are untrusted sources. This creates an indirect prompt injection surface where a malicious blog post could contain instructions designed to influence the AI agent's behavior.
Ingestion points: The scripts/scrape_blogs.py script performs HTTP GET requests to retrieve content from arbitrary URLs provided via the --urls argument.
Boundary markers: The skill does not employ boundary markers or "ignore instructions" warnings when presenting the scraped blog content (titles, descriptions, etc.) to the agent.
Capability inventory: The script is limited to network operations and console output; it does not have the ability to execute shell commands, write files, or access sensitive local directories, which significantly limits the potential impact of an injection.
Sanitization: While the script parses XML and JSON structures, it does not sanitize or filter the text content for potential malicious instructions before returning it to the agent.

blog-scraper