The Agent Skills Directory

[COMMAND_EXECUTION]: The skill patterns utilize child_process.execSync to run Node.js scripts with external input. In the provided examples, variables such as ${url} and ${outputFile} are interpolated directly into shell command strings. This practice is highly susceptible to command injection if the inputs contain shell metacharacters.\n- [PROMPT_INJECTION]: The skill creates a vulnerability surface for indirect prompt injection by ingesting untrusted data from external websites.\n
Ingestion points: Data is fetched from arbitrary URLs and saved to /tmp/output.html as defined in the 'Basic Usage Pattern'.\n
Boundary markers: The instructions do not define boundary markers or clear separations to distinguish external content from agent instructions.\n
Capability inventory: The skill uses execSync for script execution and file system access for reading/writing HTML results.\n
Sanitization: There is no mention of sanitization, validation, or escaping of the scraped HTML content before it is processed by the agent.

anti-scraping