The Agent Skills Directory

[COMMAND_EXECUTION]: The skill defines a workflow that requires the agent to execute several local Python scripts. These include schema_generator.py for creating JSON-LD markup, readability_checker.py for evaluating text friendliness for AI, visibility_tester.py for tracking brand mentions, and geo_report_generator.py for compiling quarterly reports.
[EXTERNAL_DOWNLOADS]: The visibility_tester.py script makes outbound network requests to the official APIs of well-known AI services, specifically Perplexity AI (api.perplexity.ai) and Moonshot AI's Kimi (api.moonshot.cn). These connections are used to simulate user queries and verify if a brand is being cited in AI-generated responses.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources, creating a potential surface for indirect prompt injection.
Ingestion points: Responses from external AI models are retrieved and processed within visibility_tester.py to identify brand citations.
Boundary markers: The retrieved data is parsed and stored in structured JSON files using specific keys such as context and cited, which limits the scope of the ingested text.
Capability inventory: The skill's capabilities are focused on local file manipulation (writing JSON and HTML files) and running provided Python analysis scripts.
Sanitization: The skill does not explicitly sanitize the external AI response text, as its primary function is to perform string matching for brand names and extract snippets for reporting.

geo-optimizer