The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous examples of executing the xrepo command-line utility to perform actions such as installing software, removing packages, and managing system-wide repositories.
[EXTERNAL_DOWNLOADS]: The tool is designed to download package recipes and source code from the internet. While it defaults to the vendor's repository on GitHub (xmake-io/xmake-repo), it also allows the addition of arbitrary remote repositories through the xrepo add-repo command.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection through the processing of untrusted data.
Ingestion points: The agent context may be influenced by externally provided package names, version strings, or repository URLs (e.g., in xrepo install <pkg> or xrepo add-repo <url>) processed by the commands described in SKILL.md.
Boundary markers: The skill lacks explicit boundary markers or instructions to the agent to treat package identifiers or URLs as data rather than instructions.
Capability inventory: The xrepo tool possesses significant capabilities including command execution (invoking compilers and build scripts), network access (fetching packages), and file system modification (installing/exporting packages).
Sanitization: No evidence of input validation or sanitization for user-provided package specs or URLs is present in the skill instructions.

xrepo-cli