The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it is designed to ingest and process content from a project's codebase to perform internationalization tasks.
Ingestion points: The scripts/check-linguist.mjs script recursively reads and analyzes content from a wide range of file types (JSON, TS, JS, Vue, HTML, Markdown) within the root directory of a project.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to disregard potentially malicious instructions embedded within the files it analyzes.
Capability inventory: The skill includes scripts capable of writing new files to the project structure and installing executable Git hooks.
Sanitization: Content read from the codebase is processed via simple string matching and regex without sanitization or isolation from the agent's decision-making process.
[COMMAND_EXECUTION]: The script scripts/install-linguist-hooks.sh facilitates the creation of executable files within the .git/hooks directory. These hooks are configured to trigger the execution of other scripts in the repository during standard Git workflows.
[COMMAND_EXECUTION]: The inclusion of a PowerShell wrapper (scripts/linguist-i18n.ps1) suggests execution in environments where script execution policies might be bypassed to facilitate project scaffolding.
[EXTERNAL_DOWNLOADS]: The skill documentation provides installation instructions using npx and yarn dlx that fetch content from the vendor's public GitHub repository at github.com/wundercorp/linguist-skill.

linguist