The Agent Skills Directory

[Interactive Source Confirmation]: The skill implements a strict 'STOP and confirm' policy before monitoring any external sources. It auto-discovers authoritative regulators and trackers based on user topics but requires the user to explicitly approve or modify the list before execution. This prevents the agent from autonomously visiting untrusted or unknown domains without oversight.
[URL Sanitization in Dashboard]: The dashboard generator (scripts/build_dashboard.py) includes a safe_url function that enforces a strict allowlist of protocols (http, https, mailto). This effectively mitigates the risk of URI-based attacks, such as javascript: or data: injection, when rendering links sourced from external regulatory pages.
[Restricted External Interaction]: The skill uses a domain-agnostic reputable-domain allowlist (references/sources-and-taxonomy.md) to filter any fallback web searches. It caps the number of items fetched and searches performed per run, which prevents open-ended or aggressive web crawling.
[Data Access and Privacy]: The skill utilizes workiq_get_my_profile to customize alerts for the user's specific job function. While it accesses professional profile data, it uses this information solely to flag relevant regulatory items and explicitly guards against including PII or non-public attribution in the final dashboard.
[Command Execution (Safe)]: The skill executes a local Python script (scripts/build_dashboard.py) to generate the HTML dashboard. Analysis of this script shows it uses only the Python standard library and contains no logic for sub-processes, shell execution, or remote network calls.

regulation-monitor