The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches protocol specifications, sample architectures, and SDK documentation from non-trusted external domains and GitHub repositories, including ucp.dev and github.com/Universal-Commerce-Protocol.
[REMOTE_CODE_EXECUTION]: Instructs the agent to clone the python-sdk repository and execute uv sync. This pattern involves downloading and executing environment configurations and dependencies defined in a remote repository belonging to an unverified source.
[COMMAND_EXECUTION]: The skill guides the agent to install dependencies from public registries using pip and npm. Specifically, it points to the community-maintained fastucp-python package and the @ucp-js/sdk package, which are not verified by the skill author as official.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to instructions embedded in external data it processes during the scaffolding phase.
Ingestion points: Fetches content from external READMEs, spec overviews, and web search results to guide project generation.
Boundary markers: There are no instructions or delimiters provided to ensure the agent ignores or isolates potentially malicious instructions embedded in these external sources.
Capability inventory: The skill has access to Bash (for installations), Write (for file creation), and Edit tools.
Sanitization: No validation or filtering is performed on the fetched documentation before the agent uses it to define the project structure and logic.

ucp-setup