The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches SDKs, sample code, and specification guidelines from the Universal-Commerce-Protocol GitHub repositories and official documentation domains.
[COMMAND_EXECUTION]: Instructs the agent to perform package installations using Python's pip and Node.js's npm, including a community-contributed Python package.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it directs the agent to fetch and interpret untrusted data from external websites and repositories to guide the scaffolding process.
Ingestion points: Remote URLs (including ucp.dev, github.com, and developers.google.com) are accessed via WebFetch and WebSearch to retrieve READMEs and specifications.
Boundary markers: The instructions lack explicit delimiters or warnings to the agent to treat fetched content as data rather than instructions.
Capability inventory: The skill allows for file system modifications (Write, Edit) and shell command execution (Bash).
Sanitization: There is no defined process for validating or sanitizing the content retrieved from remote sources before it influences the agent's code generation or execution steps.

ucp-setup