The Agent Skills Directory

[SAFE]: The skill demonstrates a strong security posture through its instructions and utility scripts, focusing on read-only discovery and validation of model environments.- [COMMAND_EXECUTION]: The script scripts/test_tabpfn_check.py utilizes subprocess.run to execute the skill's internal diagnostic tool, tabpfn_check.py, for unit testing purposes. The implementation uses a structured argument list and avoids shell execution, presenting no risk of command injection.- [INDIRECT_PROMPT_INJECTION]: The scripts/tabpfn_check.py utility processes user-provided JSON manifests. It implements strict schema validation, type checking, and size limits (16 MiB) to mitigate risks associated with malformed or malicious data inputs. The auditor specifically avoids outputting raw table values to prevent accidental data exposure.- [EXTERNAL_DOWNLOADS]: The documentation references external resources from established platforms including GitHub and Hugging Face for model weights and code. The instructions explicitly advise users to verify license terms manually and avoid hardcoding authentication tokens, aligning with security best practices.

tabpfn