The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on the tiangong-lca CLI tool to perform dataset operations including extract, apply, and validate. These commands are used to process JSONL files at specific file system paths provided by the user.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data which constitutes a vulnerability surface.
Ingestion points: The skill ingests data from processes.jsonl, trans-units.jsonl, source evidence documents, and process/flow QA reports.
Boundary markers: The instructions do not define clear delimiters or "ignore embedded instructions" warnings to separate user data from agent instructions during the transcreation process.
Capability inventory: The agent has the capability to execute shell commands via the tiangong-lca CLI and perform file system reads/writes on .jsonl and .json files.
Sanitization: No explicit sanitization, validation, or filtering mechanisms for the external content are described in the skill's logic.

tidas-bilingual-transcreation