The Agent Skills Directory

[SAFE]: The skill instructions explicitly prohibit the agent from performing automated actions such as sending emails, starring repositories, or posting comments, limiting its role strictly to drafting content.
[SAFE]: The skill configuration includes disable-model-invocation: true, which restricts the agent's ability to trigger additional model calls or tools, providing a controlled execution environment.
[SAFE]: A local scanning utility (generic-scan.js) is provided to validate drafts against a set of banned generic phrases, acting as a quality and safety filter.
[SAFE]: No evidence of network exfiltration, credential access, or remote code execution was found in the scripts or instructional content.

outreach