The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it fetches and processes raw HTML from untrusted external URLs provided by the user (Category 8).
Ingestion points: Website data is ingested via WebFetch as described in Phase 2a and 2b of SKILL.md.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the prompt instructions for the fetched data.
Capability inventory: The skill has the ability to execute shell commands (bash), run JavaScript via node, and execute Python scripts, primarily for report generation and validation.
Sanitization: There is no evidence of sanitization or filtering of the fetched HTML before it is processed by the agent.
[EXTERNAL_DOWNLOADS]: The skill instructions include a routine to check for and install the docx library using npm install -g docx if it is not already present. This is a standard dependency for the report generation feature.
[COMMAND_EXECUTION]: To generate the audit reports, the skill dynamically creates and executes JavaScript code using node. It also utilizes bash and python for file validation and PDF conversion (soffice.py). These operations are used for legitimate report generation and utilize specific paths within the environment.

seo-geo-aeo