The Agent Skills Directory

[SAFE]: The skill consists of prompt instructions and configuration metadata. It does not contain any executable scripts, binary files, or network-active code. The referenced scripts/sync.py is mentioned as a development tool for synchronization and is not included or executed as part of the skill's runtime instructions.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted user data (claims, conclusions, and data) for verification. While this creates a surface for indirect prompt injection, the skill employs structured reasoning and boundary markers to mitigate risks.
Ingestion points: User-provided text placeholder 【畘耴覂瀹、畐肉、数捨或方案】 in SKILL.md.
Boundary markers: The prompt uses Chinese brackets 【】 to delimit user input from instructions.
Capability inventory: The skill specifies a requirement for "networking" in the description for verification, but does not provide any script-based network operations or file-system write capabilities.
Sanitization: The skill uses a multi-step analytical framework (facts vs. conclusions vs. judgements) to force the AI to objectively deconstruct the input before reaching a final verdict.

fact-check