The Agent Skills Directory

[SAFE]: The skill prioritizes bounded execution and safety. It defines loops as feedback systems with mandatory terminal states and explicit success gates, preventing uncontrolled autonomous behavior.
[SAFE]: Instructions for handling external data are secure. The skill fetches its catalog from the author's official domain (signals.forwardfuture.ai) and explicitly instructs the agent to treat this data as untrusted reference material that cannot override core instructions or user constraints.
[SAFE]: The skill enforces human-in-the-loop requirements. It mandates explicit user approval for any destructive, irreversible, or production-impacting actions, and instructs the agent to stop if no progress is made or if a user-set limit is reached.
[SAFE]: No malicious patterns such as prompt injection, obfuscation, data exfiltration, or unauthorized privilege escalation were detected across the analyzed files.

loop-library