The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted data from external websites, creating a surface for indirect prompt injection.
Ingestion points: The scripts/seo_audit.py script fetches HTML content from user-provided URLs to extract SEO metadata.
Boundary markers: Extracted metadata such as page titles and descriptions are presented to the agent as plain text without utilizing specific delimiters or markers to prevent the agent from interpreting the content as instructions.
Capability inventory: The skill possesses the capability to execute several scripts that perform network requests and data processing.
Sanitization: While the audit script uses regular expressions for extraction and implements basic string truncation, it lacks robust sanitization to filter out potential prompt injection payloads embedded in the metadata.
[COMMAND_EXECUTION]: The skill utilizes local Python scripts and standard shell commands like curl to perform its audits and API interactions. These actions are transparently documented and aligned with the skill's primary objective of search engine optimization.

seo-geo