The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the host agent to execute a bundled Python CLI script (scripts/fetch_url.py) to perform the web scraping and extraction tasks locally.\n- [EXTERNAL_DOWNLOADS]: The skill depends on the trafilatura Python package, which users are instructed to install from PyPI. This is a well-known and established open-source library for web scraping.\n- [INDIRECT_PROMPT_INJECTION]: The skill's primary purpose is to fetch and clean content from external URLs, which creates a potential attack surface for indirect prompt injection if the source website contains instructions intended to manipulate the agent.\n
Ingestion points: The fetch_url_as_markdown function in scripts/fetch_url.py ingests data from arbitrary user-provided URLs using trafilatura.fetch_response.\n
Boundary markers: The script outputs metadata in a YAML frontmatter block, but the main content body lacks specific delimiters to isolate potentially malicious instructions from the agent's processing logic.\n
Capability inventory: While the skill itself is limited to fetching and printing content, the host agent typically possesses additional tools (file access, terminal access) that could be targeted via injection.\n
Sanitization: The script performs structural extraction to provide clean Markdown but does not include sanitization logic to detect or filter instructions targeted at LLMs within the page content.

fetch-url-as-markdown