The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the tiangong-lca CLI, npm, and node to manage dataset orchestration, curation queues, and patch application. These operations are standard for the skill's documented purpose of data import and processing.
[UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The skill resolves and executes additional logic from the vendor's repository (https://github.com/tiangong-ai/skills) at runtime using the npx skills command. This is used to incorporate specific research capabilities (e.g., academic search) into the authoring workflow.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted external data, including PDFs, Excel files, and academic journal papers, to generate structured TIDAS data using AI.
Ingestion points: Processes source files from local paths and academic evidence retrieved via auxiliary research skills.
Boundary markers: The skill includes a 'Boundaries' section in SKILL.md that restricts the scope of the agent's orchestration and behavior.
Capability inventory: The skill possesses executable capabilities including CLI execution (tiangong-lca, npx), local script execution (node, npm), and workspace file writing.
Sanitization: Implements a series of downstream gates, including schema validation, deterministic QA classification, and curation gate reviews, to verify AI-authored data before it is finalized.

tidas-data-import