The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides several Python scripts (analyze.py, http_checker.py, internal_links.py, link_graph.py, outbound_links.py) designed to be executed via a shell to process files in a user-provided distribution directory. These scripts perform local file system operations and site-wide auditing as part of the intended workflow.
[DATA_EXFILTRATION]: The http_checker.py script performs network operations (HTTP HEAD and GET) to external, non-whitelisted domains to validate link status. Because this activity is the primary purpose of the skill and does not involve accessing or transmitting sensitive system context, the risk is categorized as minimal.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data by parsing HTML files from a user-specified directory to extract link attributes. While it lacks explicit delimiters to ignore embedded instructions, it only extracts structured link data and does not interpolate the content into agent instructions or execute it.
Ingestion points: HTML content read from the local file system during analysis phases in scripts/outbound_links.py, scripts/internal_links.py, and scripts/link_graph.py.
Boundary markers: None present.
Capability inventory: The skill has the ability to read any file in the provided directory, write analysis reports to disk, and perform network requests via the requests library.
Sanitization: Includes URL normalization and specific status code filtering for known bot-blocking domains (e.g., LinkedIn, Twitter) to handle false positives.

link-analyzer