The Agent Skills Directory

[SAFE]: The skill consists exclusively of documentation, configuration files, and metadata. There are no scripts, binaries, or automated shell commands included that could execute on the host system.
[EXTERNAL_DOWNLOADS]: The skill references documentation and asset URLs from commercengine.io, github.com/commercengine, and tarkai.com. These are official resources belonging to the skill author and are used for documentation, source code hosting, and branding (logo).
[INDIRECT_PROMPT_INJECTION]: The skill documentation outlines how an agent should handle user requests for orders, payments, and shipments. While this defines an ingestion surface for user-provided data, the skill does not implement any logic that would unsafely interpolate this data into sensitive operations.
Ingestion points: User requests such as "Place order", "Order history", and "Track shipment" as described in the Decision Tree in SKILL.md.
Boundary markers: Not present in the provided SDK usage examples.
Capability inventory: The frontmatter specifies Bash in allowed-tools, but the skill package contains no Bash scripts.
Sanitization: The documentation assumes the use of the sdk.order library, which is responsible for validating and sanitizing inputs such as order_number and cart_id.

ce-orders