The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits a susceptibility to indirect prompt injection as it fetches and processes data from untrusted external websites (competitors and third-party directories) using the WebFetch tool.
Ingestion points: The skill uses WebFetch in Step 3 (NAP check) and Step 6 (Competitor Benchmarking) to extract content from various external URLs.
Boundary markers: The instructions lack specific delimiters or instructions to ignore embedded commands within the fetched external data.
Capability inventory: The skill has access to Bash, WebFetch, and Read tools.
Sanitization: There is no defined process for sanitizing or validating the content retrieved from external websites.
[PROMPT_INJECTION]: The skill employs context modification by instructing the agent to operate under the assumption that the current date is March 2026, referencing 'future' Google playbooks and research to guide SEO strategy.
[COMMAND_EXECUTION]: The skill requests the Bash tool in its configuration. While no specific executable scripts or dangerous shell commands are defined within the text, the presence of a shell tool increases the potential attack surface if an indirect injection were to occur.
[EXTERNAL_DOWNLOADS]: The skill fetches data from well-known technology platforms and services, including Yelp, Facebook, Apple Maps, and Bing, to perform NAP (Name, Address, Phone) consistency audits and competitor benchmarking.

30x-seo-local