The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external data from Amazon listings, including product titles, bullet points, and descriptions, as well as search query performance data.
Ingestion points: Untrusted data is retrieved from Amazon catalog and performance sources using MCP tools such as amazon_listings_raw and amazon_child_product_organic_search_ranks_per_week (referenced in SKILL.md).
Boundary markers: The instructions do not define specific delimiters or instructions to prevent the agent from following directions that might be embedded within the listing copy being optimized.
Capability inventory: The skill primarily involves data reading and report generation; while it mentions a separate write skill (AMAZON_LISTINGS_UPDATE), the current instructions are read-only.
Sanitization: No explicit sanitization or filtering of the retrieved listing content is mandated before analysis. This represents a potential surface for indirect injection if an attacker controls a listing's content, though this is a low-risk concern inherent to the skill's purpose.
[DATA_EXFILTRATION]: The skill configuration specifies a remote Model Context Protocol (MCP) base URL at mcp.datadoe.com.
Network operations: The agent is instructed to interact with this non-whitelisted domain to fetch listing and search data.
Context: As this domain provides the core infrastructure for the skill's functionality, these network operations are considered legitimate within the scope of the intended use case.

amazon-listing-optimizer