The Agent Skills Directory

[SAFE]: The skill implements strong cookie security by using add_unredirected_header for the session cookies, ensuring sensitive authentication data is not leaked to third-party domains if a request is redirected. It also scopes cookie usage to specific domains.
[SAFE]: The image rehosting feature includes multiple layers of SSRF protection. It validates that image URLs are public, resolves hostnames to verify IP ranges (blocking private, loopback, and reserved IPs), and uses a custom redirect handler to prevent bypasses via HTTP redirects.
[SAFE]: Destructive or state-changing actions, such as publishing a story, are gated behind a mandatory --confirm flag. The instructions explicitly direct the agent to perform a dry-run and obtain user consent before execution.
[SAFE]: The skill uses only standard Python libraries (urllib, ipaddress, argparse) and does not introduce any third-party dependencies or remote code execution vectors.

medium