The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted HTML data, which serves as an ingestion point for external instructions. * Ingestion points: HTML input is processed via the html-to-markdown CLI tool and the Python convert function. * Boundary markers: The skill does not provide delimiters or instructions to the agent to ignore potentially malicious commands embedded in the processed HTML. * Capability inventory: The skill utilizes shell command execution (html-to-markdown, jq) and Python script execution. * Sanitization: No sanitization or validation of the extracted metadata is mentioned or demonstrated.
[COMMAND_EXECUTION]: The skill provides instructions for the agent to execute shell commands using html-to-markdown and jq to transform and filter data.

extracting-metadata