The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines workflows that require the agent to analyze the existing codebase and translation files to identify patterns and implement new features. This creates a surface for indirect prompt injection if the translation files or source code ingested by the agent contain malicious instructions designed to influence the agent's behavior.
Ingestion points: Workflow instructions in SKILL.md requiring the agent to explore-codebase and analyze translation patterns, as well as the runtime processing of JSON locale files via the t() function and Trans component.
Boundary markers: The skill relies on standard React and i18next safeguards; the i18n/config.ts template uses interpolation: { escapeValue: false } because React provides built-in XSS protection for interpolated strings.
Capability inventory: The agent has the ability to read the codebase, perform research via external documentation tools (Exa/Context7), and modify application files.
Sanitization: The skill employs standard React DOM escaping to handle dynamic content within translations.

react-i18n