The Agent Skills Directory

[SAFE]: The skill serves as a high-level controller for a mathematical modeling workflow. Its primary function is project organization and coordination of downstream skills.
[COMMAND_EXECUTION]: Although the skill is granted Bash(*) permissions, the instructions are limited to standard file system operations (creating and updating markdown files) within the local project directory. No suspicious shell commands, system modifications, or privilege escalation patterns were detected.
[DATA_EXFILTRATION]: No hardcoded credentials or attempts to access sensitive system files (e.g., SSH keys, AWS configs) were found. The use of network-enabled tools like WebSearch and WebFetch is consistent with the stated purpose of research for modeling competitions.
[PROMPT_INJECTION]: The instructional language is structured and follows a logical workflow. It does not contain attempts to override agent safety guidelines, bypass system constraints, or extract internal configuration.
[EXTERNAL_DOWNLOADS]: The skill does not perform any remote code downloads or execute unverified external scripts. It relies on a local reference file (math_modeling_norms.md) for domain knowledge.

1start-mathmodel