The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes data from external, untrusted websites during SEO audits.
Ingestion points: Data retrieved from external URLs by lighthouse.sh and API responses from pagespeed.sh.
Boundary markers: None identified in the provided instructions to delimit external content from agent instructions.
Capability inventory: The skill uses shell execution (sh, mjs) and network operations (curl) which could be influenced by injected content.
Sanitization: No sanitization or validation of the external report content is mentioned.
[COMMAND_EXECUTION]: The skill facilitates the execution of shell scripts (scripts/lighthouse.sh, scripts/pagespeed.sh) and JavaScript files (scripts/search-console-export.mjs). These scripts are used to automate the auditing process and interact with local and remote systems.
[EXTERNAL_DOWNLOADS]: The skill includes instructions to download its components from a remote GitHub repository. Furthermore, the provided scripts perform network operations to fetch data from the PageSpeed Insights API and external websites for Lighthouse analysis.
[CREDENTIALS_UNSAFE]: The documentation encourages users to pass sensitive information, such as PAGESPEED_API_KEY and GSC_ACCESS_TOKEN, directly as environment variables in shell commands. This practice can lead to credentials being inadvertently stored in shell history files.

seo-aeo-audit