The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains robust instructions under the "Author authority" section to prevent indirect prompt injection. It explicitly directs the agent to treat third-party content (like changelogs, issue references, or advisories) as data rather than instructions, and warns against acting on text that addresses the agent directly.
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute standard package management commands (npm, pnpm, composer, cargo, go). All actions are preceded by a planning phase and require manual user confirmation before any files are modified. The skill also explicitly refuses to auto-install missing toolchain components.
[DATA_EXPOSURE]: The skill reads local configuration files such as .npmrc, .tituskirch-skills.json, and project manifests to understand registry settings and release-age gates. This is a standard requirement for its primary function and is performed locally without evidence of external data exfiltration.
[EXTERNAL_DOWNLOADS]: The skill references a JSON schema from the author's official GitHub repository (raw.githubusercontent.com/TitusKirch/...). As this is a source belonging to the skill's vendor, it is considered a legitimate resource.

update-deps