The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses localized Python scripts (stencil_generator.py, scheme_selector.py, truncation_error.py) to perform mathematical computations. These scripts use argparse for strict input validation, including bounds checking on derivative order (max 6), accuracy (max 8/12), and stencil offsets (max 51). Subprocess calls from the agent to these scripts use explicit argument lists without shell interpolation.
[REMOTE_CODE_EXECUTION]: No remote code execution patterns were found. The skill does not use curl, wget, pip, or any other network-related tools. It operates entirely on local scripts and reference files.
[DATA_EXFILTRATION]: No network access or sensitive file access is requested or performed. The scripts only read from CLI arguments and write to standard output.
[PROMPT_INJECTION]: The instructions are focused on mathematical workflows. There are no attempts to override agent behavior, bypass safety filters, or extract system prompts. The 'Verification checklist' and 'Common pitfalls' sections in SKILL.md actually serve as defensive measures to ensure the AI provides accurate mathematical advice.
[OBFUSCATION]: No obfuscated content, Base64 strings, or hidden characters were detected in the scripts or documentation.
[INDIRECT_PROMPT_INJECTION]: The skill processes mathematical parameters (integers/floats) for numerical schemes. It does not ingest untrusted data from external sources like URLs or arbitrary user-controlled files that could contain hidden instructions.

differentiation-schemes