The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external release notes and changelogs to generate summaries and decide on update applications. This introduces a vulnerability where malicious instructions embedded in a package's changelog could attempt to influence the agent's behavior or file modifications.
Ingestion points: External changelogs and release notes (SKILL.md, "Analyze release impact" section).
Boundary markers: None specified; the agent is instructed to aggregate findings without explicit instructions to ignore potentially malicious content within the external text.
Capability inventory: File system write access (in "apply mode"), command execution (vpx taze, vp update, and standard ecosystem updaters).
Sanitization: No specific sanitization or filtering logic is defined for the aggregated external text.
[COMMAND_EXECUTION]: The skill instructs the agent to execute several command-line tools including vpx taze, vp update, and various ecosystem-standard package managers. These tools are used to query for updates and modify the repository environment.
[EXTERNAL_DOWNLOADS]: The skill fetches version information from the official Node.js release index at https://nodejs.org/dist/index.json. This is a well-known service used for legitimate version comparison.

repository-updater