The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill resolves and utilizes external skills at runtime via npx skills use targeting the vendor's repository at https://github.com/tiangong-ai/skills. This is used to incorporate specialized research capabilities like tiangong-kb-sci-search as described in the Source Document Authoring lane.
[REMOTE_CODE_EXECUTION]: The skill employs npx skills use to download and execute code from the vendor's remote GitHub repository. While this represents a remote code execution pattern, the source is within the author's official namespace and the purpose is explicitly defined for academic research retrieval.
[COMMAND_EXECUTION]: The instructions rely on a series of shell commands and scripts to manage the data lifecycle, including tiangong-lca CLI invocations for dataset authoring and conversion, as well as npm and node script calls for gating and finalization.
[INDIRECT_PROMPT_INJECTION]: The skill processes external, potentially untrusted documents such as PDFs, Excel files, and web exports to generate candidate data.
Ingestion points: Path arguments provided to tiangong-lca dataset author and tiangong-lca dataset import-lca in SKILL.md.
Boundary markers: The agent is instructed to avoid retaining __AI_FILL_* placeholders and to verify outputs against structured report artifacts.
Capability inventory: The skill can execute CLI tools, run Node.js scripts, and resolve remote skills via npx.
Sanitization: The workflow includes multiple verification gates, including SDK validation, QA classification scripts, and a finalize wrapper that reruns rulesets and generates mutation manifests before any remote write occurs.

tidas-data-import