The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to fetch protocol specifications from ucp.dev and a conformance test suite from the Universal-Commerce-Protocol repository on GitHub. These downloads are used for documentation and implementation verification purposes.
[PROMPT_INJECTION]: The skill describes a platform architecture that ingests external data via webhooks. This creates a surface for indirect prompt injection (Category 8).
Ingestion points: Webhook endpoint POST body (SKILL.md).
Boundary markers: None explicitly defined for isolating data from agent instructions.
Capability inventory: The skill allows powerful tools including Bash, Write, Edit, and WebFetch (SKILL.md).
Sanitization: The skill mandates a robust authentication mechanism using Detached JWT signatures (RFC 7797) verified against public keys found in the business's /.well-known/ucp profile. While this ensures data authenticity and integrity, it does not sanitize the contents of the order data against prompt injection if processed by an LLM.
[COMMAND_EXECUTION]: The skill requests access to the Bash tool. While no malicious shell commands are provided in the instructions, the presence of this tool alongside the ingestion of external data requires careful monitoring during execution to prevent command injection.

ucp-orders-webhooks