The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill scans untrusted repository source code and documentation to detect translatable strings and locale hints, creating a surface for indirect prompt injection.
Ingestion points: Phase 1 detection (described in references/detect/js-ts.md) reads the sourceDir and README files.
Boundary markers: The skill does not implement explicit delimiters or boundary markers when passing detected code snippets to the agent for analysis.
Capability inventory: The skill can execute shell commands via the crowdin CLI, write repository files, and delegate tasks to external agents.
Sanitization: Ingested content from source files is not filtered or sanitized before processing.
[COMMAND_EXECUTION]: The skill utilizes the crowdin CLI and git commands to manage internationalization workflows, project configurations, and repository synchronization. These commands are necessary for the skill's primary functionality and are used according to standard practices.
[EXTERNAL_DOWNLOADS]: The skill facilitates the installation of the @crowdin/cli package and delegates specialized tasks to agents from the Lingui ecosystem (github.com/lingui/skills). These downloads and delegations target well-known and reputable services essential to the skill's operations.

i18n-setup