The Agent Skills Directory

[COMMAND_EXECUTION]: The skill facilitates the execution of local Python tools including huashu_lint.py for compliance scanning and atoms_query.py for database retrieval during the content generation workflow.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process brand-specific information and local marketing templates to produce conversational output, creating a potential surface for indirect prompt injection.
Ingestion points: Business information inputs (brand, category, scene) and the assets/真实语料-待填.md corpus file.
Boundary markers: None explicitly defined in the provided instruction set.
Capability inventory: Execution of local validation scripts and knowledge base queries.
Sanitization: Employs a mandatory compliance scanner that checks for prohibited advertising claims and unauthorized data collection requests.
[DYNAMIC_EXECUTION]: The scripts/huashu_lint.py utility dynamically calculates the project root and modifies the Python environment path at runtime to load shared libraries, which is a method for dynamic code loading from computed paths.

siyu-huashu