The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a set of shell commands in references/checks.md intended for website auditing. These commands utilize standard utilities including curl, grep, sed, dig, and python3 (specifically the json.tool module) to process site metadata and structure.
[EXTERNAL_DOWNLOADS]: The audit workflow involves fetching data from remote websites using curl. This includes retrieving robots.txt, XML sitemaps, and page HTML to verify indexability and machine-readability.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external websites, which represents a potential attack surface for indirect prompt injection.
Ingestion points: The skill fetches external website content, HTTP headers, and JSON-LD structured data via curl as defined in references/checks.md.
Boundary markers: The provided scripts do not explicitly define delimiters to separate fetched content from agent instructions, though this is typical for modular diagnostic tools.
Capability inventory: The skill utilizes shell execution for network requests (curl, dig) and text processing (grep, sed, python3).
Sanitization: The skill employs grep -oE and sed to isolate specific HTML tags and scripts, providing a level of filtering for the ingested data.

ai-visibility-audit