The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill utilizes the WebFetch tool to retrieve content from external URLs provided by the user, including homepages and sitemaps. This activity is central to the skill's functionality but exposes the agent to data from untrusted origins.
[COMMAND_EXECUTION]: The skill requests access to the Bash tool. While no specific malicious commands are hardcoded in the workflow, the availability of a shell represents a significant capability that could be exploited if the agent is misled by malicious input.
[PROMPT_INJECTION]: The skill is susceptible to Indirect Prompt Injection (Category 8) due to its core function of processing untrusted web content.
Ingestion points: External website content (HTML, metadata, schema) fetched via WebFetch in Phase 1.
Boundary markers: Absent. The instructions to the subagents in Phase 2 do not define boundaries or instruction-ignore tags for the content being analyzed.
Capability inventory: Bash, Write, Read, Grep, Glob, WebFetch.
Sanitization: Absent. There is no mention of filtering or validating the fetched content for injection patterns before analysis.

geo-audit