The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a shell script scripts/spreadsheet_workflows.sh used for environment verification. It checks for the presence of necessary system binaries (python3, uv, soffice, pdftoppm) and Python modules. These operations are restricted to local environment discovery.
[EXTERNAL_DOWNLOADS]: The skill identifies dependencies on well-known and trusted Python packages such as openpyxl, pandas, and matplotlib. It provides deterministic installation commands (e.g., via uv pip install) without executing them automatically, allowing for user-mediated setup.
[REMOTE_CODE_EXECUTION]: No patterns of arbitrary remote code execution were detected. The skill instructions and scripts focus on local file manipulation using standard libraries.
[PROMPT_INJECTION]: The skill includes explicit security testing in references/evals.yaml to ensure the agent resists attempts to bypass instructions or execute unauthorized commands like rm -rf or curl. This demonstrates an intentional focus on robustness against adversarial inputs.
[DATA_EXFILTRATION]: No evidence of unauthorized data transfer or access to sensitive local files (such as SSH keys or cloud credentials) was found. The skill operates within designated temporary and output directories.

spreadsheet