The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external data and reports to generate executable code, which presents an attack surface for instructions embedded in source data.
Ingestion points: The agent reads reports/ANALYSIS_MODELING_REPORT.md, ../mathmodel-references/math_modeling_norms.md, and raw project datasets to guide code generation.
Boundary markers: Absent. The instructions do not define specific delimiters to separate untrusted data from instructions.
Capability inventory: The agent uses Bash(*), Write, Edit, and WebFetch tools to execute logic and fetch external data based on the content of these inputs.
Sanitization: Absent. The skill does not provide mechanisms to sanitize or validate the content of the modeling reports or datasets before they are used to generate code.
[DYNAMIC_EXECUTION]: The skill's primary function is the runtime generation and execution of scripts.
The instructions mandate that the agent "编写可复现代码" (write reproducible code) and "运行求解" (run solution) within the code/ directory.
Scripts are executed using the Bash(*) tool, allowing for arbitrary command execution as determined by the agent's interpretation of the modeling task.

3coding-visual