The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute dynamically generated Python scripts (python3 -c) for core logic, including data deduplication, translation tracking, and merging batch results.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection (Category 8) as it ingests over 3,000 questions from external JSON files and passes them to sub-agents for translation without sanitization.
Ingestion points: Question (q) and answer (a, b, c) fields in src/data/*.json files.
Boundary markers: The sub-agent prompt template lacks delimiters or instructions to ignore commands embedded within the data.
Capability inventory: The skill has access to Write, Bash, and Task tools, which could be abused if the agent obeys instructions found within the driving questions.
Sanitization: No filtering or escaping is applied to the ingested text before it is processed by the LLM or written to files.
[COMMAND_EXECUTION]: Step 5 involves modifying application source code (src/js/i18n.js, src/index.html, etc.) based on a user-supplied language variable. Without strict validation of the {lang} input, this presents a risk of code injection into the web application's frontend files.

translate-questions