The Agent Skills Directory

[SAFE]: The skill utilizes a dedicated Python utility (scripts/csv-library.py) to manage local CSV libraries. The script ensures data integrity through strict schema enforcement, validating fixed column counts and header names, and performs atomic file updates using temporary files and os.replace to prevent data corruption.
[SAFE]: The workflow is explicitly restricted to local operations. No network capabilities, external resource downloads, or remote code execution patterns were identified in the skill instructions or scripts.
[SAFE]: The skill design follows the principle of least privilege and user control by mandating interaction gates (via the AskUserQuestion tool) as confirmation steps before any file initialization, data import, or library mutation.
[INDIRECT_PROMPT_INJECTION]: The skill features a data ingestion surface by processing local CSV and JSON files.
Ingestion points: Processes the project's product-library.csv and imports external data from user-supplied CSV and JSON files via the helper script.
Boundary markers: Instructions require the agent to present an approval gate (confirmation) before performing imports or writes.
Capability inventory: The skill uses Bash to invoke the local Python helper and Write for file system modifications.
Sanitization: The Python helper script strictly validates CSV structure, row lengths, and UTF-8 encoding, failing safely if any content does not match the expected canonical schema, which mitigates the risk of processing malicious data payloads.

master-schedule